2019_Vandera_Kelli_Kalliopi_1... - King's Research Portal

This electronic thesis or dissertation has been

downloaded from the King’s Research Portal at

https://kclpure.kcl.ac.uk/portal/

Take down policy

If you believe that this document breaches copyright please contact [email protected] providing

details, and we will remove access to the work immediately and investigate your claim.

END USER LICENCE AGREEMENT

Unless another licence is stated on the immediately following page this work is licensed

under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

licence. https://creativecommons.org/licenses/by-nc-nd/4.0/

You are free to copy, distribute and transmit the work

Under the following conditions:

Attribution: You must attribute the work in the manner specified by the author (but not in anyway that suggests that they endorse you or your use of the work).

Non Commercial: You may not use this work for commercial purposes.

No Derivative Works - You may not alter, transform, or build upon this work.

Any of these conditions can be waived if you receive permission from the author. Your fair dealings and

other rights are in no way affected by the above.

The copyright of this thesis rests with the author and no quotation from it or information derived from it

may be published without proper acknowledgement.

A biophysical investigation of a membrane-active cyclodextrin-inliposome formulationfor antibiotic delivery

Vandera, Kalliopi-Kelli Apostolos

Awarding institution:King's College London

Download date: 06. Jul. 2022

0

A biophysical investigation of a membrane-active cyclodextrin-in-liposome formulation for antibiotic

delivery

A thesis submitted by

Kelly-Kalliopi Vandera

In fulfilment of the requirement for the degree of Doctor of Philosophy in

the Institute of Pharmaceutical Science of the School of Biomedical and

Life Sciences

2018

Department of Pharmacy

King’s College London

October, 2018

i

Acknowledgements

First, I would like to express my deepest gratitude to Dr. Richard Harvey and Dr. Cecile Dreiss

for imparting their knowledge and offering their invaluable support and guidance throughout

my research. They are both exceptional supervisors and friends! I could not have asked for

better supervisors. I truly appreciate their help and encouragement!

I would like to thank Dr. Stuart Jones for believing in me during my MSc research project and

introducing me to Richard as my PhD supervisor. A further big thank you to Dr. Miraz Rahman

for his warm welcome into his lab, as well as Dr. Kazi Nahar and Dr. Pietro Picconi for teaching

me drug synthesis. I would also like to thank Dr. Andrew Chan for enlightening me with his

expertise; and Dr. Ken Bruce for both letting me work in the microbiology lab and his advice

on how to write a thesis. I greatly appreciate the assistance of Dr. Luke Clifton and Dr.

Maximilian Skoda during my neutron time at the ISIS facilities. A big thanks goes to Professor

Peter Hylands and Nicola Tingley for their assistance when I was “banned” accidentally from

KCL (funny, but it is true!).

I acknowledge all the people that helped me throughout my PhD: Dr. Gianluca Bello for giving

me advice and tips on Langmuir trough, when I was really struggling to find answers to things

that didn’t make any sense; Dr. Helene Marbach, Dr. Masirah Zain, Dr. Arcadia Woods, Dr.

Giorgia Manzo, Dr. Margarita Valero and Dr. Raquel Barbosa for their invaluable help

concerning microbiology and formulation techniques; and Simona Di Blasio, DeDe and

Caroline thank you for your advice and your support. I am really grateful to you all for making

my PhD life more enjoyable, full of smiles and good vibes.

Last but not least, I would like to thank my family for their endless support. I am exceptionally

grateful to my late father, Apostole Vanderas, for providing me with the ethical foundation,

strength and patience to achieve my life goals, during the first 17 years of my life, and my

mother, Sophia Dimeli, for being both a mother and a father up to now. Mum, thank you from

the bottom of my heart for being my mainstay throughout the PhD. Chryssanthi Vandera, thank

you for being such a supportive sister and stand beside me at every step of my PhD. I would

also like to thank my in-laws, John and Katie Gateley, for supporting me through this journey.

A special thanks goes to my beloved husband, Dimitri Gateley, for believing in me, supporting

and encouraging me over the last 4 years. I am grateful for having such an affectionate

husband, who loves me unconditionally and makes me smile.

Thank you all for being part of this journey!

List of publications

ii

List of publications Hubbard ATM, Barker R, Rehal R, Vandera K-KA, Harvey RD, Coates ARM. Mechanism of

Action of a Membrane-Active Quinoline-Based Antimicrobial on Natural and Model Bacterial

Membranes. Biochemistry. 2017;56(8):1163-1174.

Kumar A, Terakosolphan W, Hassoun M, Vandera K-KA, Novicky A, Harvey RD, Royall PG,

Bicer EM, Eriksson J, Edwards K, Valkenborg D, Nelissen I, Hassall D, Mudway IS, Forbes B.

A Biocompatible Synthetic Lung Fluid Based on Human Respiratory Tract Lining Fluid

Composition. Pharm Res. 2017;34(12):2454-2465.

Urbano L, Clifton L, Ku HK, Kendall-Troughton H, Vandera K-KA, Matarese BFE, Abelha T, Li

P, Desai T, Dreiss CA, Barker R, Green M, Dailey LA, Harvey RD. Influence of the Surfactant

Structure on Photoluminescent π-Conjugated Polymer Nanoparticles: Interfacial Properties

and Protein Binding. Langmuir. 2018;34(21):6125-6137.

Abstract

iii

Abstract Gram-negative bacteria possess numerous defence mechanisms against antibiotics, due to

the intrinsic permeability barrier afforded by their outer membrane and the various efflux

mechanisms which pump out drugs, explaining the recalcitrance of some common and life-

threatening infections. A novel compound, PPA148, was synthesised in-house and showed

promising activity against Gram-negative bacteria. Nevertheless, in some clinical bacterial

strains, drug efflux resulted in reduced efficacy was observed, which was reversed in the

presence of the efflux pump inhibitor phenylalanine-arginine β naphthylamide (PAβN). A

formulation of PPA148 consisting of a drug/cyclodextrin (drug/CD) complex encapsulated in

liposomes is investigated as a way to increase drug uptake. PPA148 has very low water

solubility, which was measured using spectroscopic techniques: photon correlation

spectroscopy (PCS) and Ultra-Violet spectroscopy (UV). Its solubility is improved by

complexation with βCD derivatives (HPβCD and RAMEB) and the complex is characterized

by applying nuclear magnetic resonance (NMR) and fluorescence. Fluidosomes, loaded with

the drug/CD complex, are manufactured by applying the thin-film hydration method followed

by extrusion for reducing the size of liposomes. PCS is used to size the particles and measure

their zeta-potential. A Stewart assay and UV were used to quantify the lipid and drug

concentration respectively, in the final formulation. A disk diffusion microbiological assay is

used to assess the efficacy of the formulation against E. coli (DH5a). A variety of in vitro

biophysical techniques are used to assess the mechanism of the drug uptake. Phospholipids,

Re Lipid A extracted from S. Minnesota and Rc J5 LPS extracted from E. coli are used to

make a synthetic monolayer and bilayer model of the outer and inner bacterial membranes.

They are adapted for use with the Langmuir trough (LT) and neutron reflectivity (NR)

techniques to monitor changes occurring upon interaction with the drug alone and the

formulated drug. The results revealed an possible increase in drug efficacy with a possible

fusion mechanism of uptake. NR provides a new method to examine the fusion mechanism of

fluidosomes with bacterial membrane.

Table of Contents

iv

Table of Contents Acknowledgements ........................................................................................................................................ i

List of publications ......................................................................................................................................... ii

Abstract ........................................................................................................................................................ iii

Table of Contents .......................................................................................................................................... iv

Table of Figures ........................................................................................................................................... viii

Table of Tables ............................................................................................................................................. xv

Abbreviations ............................................................................................................................................. xvii

Symbols ...................................................................................................................................................... xix

Physical Constants ...................................................................................................................................... xxi

Chapter 1 General Introduction .............................................................................................................. 1

1.1 Antibiotics and Bacteria: a unique class of medicines and pathogens ................................... 2

1.2 Antibiotic classification ........................................................................................................ 2

1.3 Bacterial classification .......................................................................................................... 3

1.3.1 Physicochemical properties of antibiotics ........................................................................................... 3

1.3.2 Gram negative bacterial cell envelope as a barrier to antibiotics ....................................................... 4

1.3.3 Drug uptake mechanisms .................................................................................................................... 6

1.4 Antimicrobial Resistance (AMR) ........................................................................................... 7

1.4.1 Causes of AMR ..................................................................................................................................... 8

1.4.2 Mechanisms of bacterial resistance .................................................................................................... 9

1.5 Strategies to combat antimicrobial resistance .................................................................... 10

1.5.1 Drug development ............................................................................................................................. 11

1.5.2 Antibiotic combination therapy ........................................................................................................ 11

Table of Contents

v

1.6 Formulation approaches for antimicrobials to overcome resistance ................................... 13

1.6.1 Cyclodextrin-drug complexes ............................................................................................................ 13

1.6.2 Liposomal carriers ............................................................................................................................. 16

1.7 Aims and scope of this project ............................................................................................ 17

1.7.1 Drug-in-Cyclodextrin-in-Liposomes ................................................................................................... 18

1.7.2 Plan of the study and techniques used ............................................................................................. 19

1.7.3 Interfacial techniques used to investigate drug-membrane interactions ......................................... 20

Chapter 2 Synthesis, characterization and solubility of PPA148 .............................................................. 28

2.1 Introduction ....................................................................................................................... 29

2.2 Materials ............................................................................................................................ 32

2.3 Methods ............................................................................................................................ 33

2.3.1 Synthetic pathway of PPA148 ........................................................................................................... 33

2.3.2 Thin Layer Chromatography (TLC) ..................................................................................................... 37

2.3.3 Liquid Chromatography-Mass Spectrometry (LC/MS) ....................................................................... 37

2.3.4 Nuclear Magnetic Resonance (NMR) spectroscopy .......................................................................... 38

2.3.5 Fourier Transform Infra-Red (FTIR) Spectroscopy ............................................................................. 38

2.3.6 Adsorption assay ............................................................................................................................... 38

2.3.7 Ultra Violet Spectroscopy .................................................................................................................. 39

2.3.8 Turbidimetric Assay ........................................................................................................................... 40

2.3.9 Water Solubility Test ......................................................................................................................... 40

2.3.10 Quantification of drug:cyclodextrin binding constant by fluorescence spectroscopy .................... 41

2.3.11 Phase solubility diagram .................................................................................................................. 41

2.3.12 Continuous method variation (Job’s plot) (156) .............................................................................. 43

2.4 Results ............................................................................................................................... 44

Table of Contents

vi

2.4.1 Synthesis ............................................................................................................................................ 44

2.4.2 Physicochemical characteristics ........................................................................................................ 50

2.5 Discussion .......................................................................................................................... 58

2.6 Conclusion ......................................................................................................................... 62

Chapter 3 Liposomaly encapsulated drug/cyclodextrin inclusion complexes .......................................... 63

3.1 Introduction ....................................................................................................................... 64

3.2 Materials ............................................................................................................................ 66

3.3 Methods ............................................................................................................................ 66

3.3.1 Formation of Large Multi-Lamellar Fluidosomes .............................................................................. 66

3.3.2 Size reduction methods for the manufacture of Fluidosomes .......................................................... 67

3.3.3 Preparation of drug-cyclodextrin inclusion complexes ..................................................................... 67

3.3.4 Incorporation of drug/cyclodextrin complexes into liposomes ........................................................ 67

3.3.5 Size Exclusion Chromatography ........................................................................................................ 67

3.3.6 Photon Correlation Spectroscopy (PCS) ............................................................................................ 68

3.3.7 Stewart assay ..................................................................................................................................... 68

3.3.8 Drug quantification ............................................................................................................................ 69

3.3.9 Kirby Bauer assay .............................................................................................................................. 70

3.3.10 Dispersion stability of empty fluidosomes ...................................................................................... 71

3.3.11 Statistical analysis ............................................................................................................................ 71

3.4 Results ............................................................................................................................... 71

3.4.1 Characterization of empty/control Fluidosomes ............................................................................... 71

3.4.2 Characterization of drug-CD inclusion complexes in the Fluidosomes ............................................. 75

3.4.3 Encapsulation Efficiency and Drug Loading ....................................................................................... 76

3.4.4 Kirby Bauer assay .............................................................................................................................. 77

Table of Figures

vii

3.5 Discussion .......................................................................................................................... 80

3.6 Conclusion ......................................................................................................................... 84

Chapter 4 A biophysical investigation into the uptake mechanism of PPA148 and its delivery system .... 86

4.1 Introduction ....................................................................................................................... 87

4.2 Materials ............................................................................................................................ 90

4.3 Methods ............................................................................................................................ 90

4.3.1 Surface Pressure-Area isotherms at the air-liquid interface ............................................................. 90

4.3.2 Drug - monolayer interaction ............................................................................................................ 91

4.3.3 Drug - lipid bilayer interaction using fluorescence spectroscopy ...................................................... 92

4.3.4 Neutron reflectivity of asymmetric bilayer of DPPC and LPS and interaction with fluidosomes ...... 93

4.4 Results ............................................................................................................................... 97

4.4.1 Surface Pressure-Area isotherms of Langmuir monolayers at the air-liquid interface ..................... 97

4.4.2 Drug interactions with model inner membranes ............................................................................ 101

4.4.3 Drug interaction with model outer membranes ............................................................................. 107

4.5 Discussion ......................................................................................................................... 115

4.6 Conclusion ........................................................................................................................ 119

Chapter 5 General Conclusion .............................................................................................................. 120

5.1.1 Future Work .................................................................................................................................... 123

References ................................................................................................................................................. 125

Appendix A ................................................................................................................................................. 141

Quantum-mechanic rule ......................................................................................................... 141

Appendix B ................................................................................................................................................. 142

Appendix C ................................................................................................................................................. 145

Table of Figures

viii

Table of Figures Figure 1.1: Mechanism of action of some major antibiotic agents reproduced from Clatworthy et al. (10).

............................................................................................................................................................... 3

Figure 1.2: Cell envelope of Gram-negative bacteria reproduced from Silhavy et al. (18). ................... 5

Figure 1.3: Structure of phospholipid headgroups with different substitution in the phosphate group which gives a different curvature and charge. R1 and R2 are the fatty acid chains which can be either

saturated (PC with C16:0 is DPPC while PC with C14:0 is DMPC) or unsaturated (PC with C16:1 is

POPC). ................................................................................................................................................... 6

Figure 1.4: Different ways by which a substance can be transported across the membranes reproduced from Bolla et al., (34). Hydrophilic components and nutrients can pass through porins (a) and

hydrophobic molecules diffuse through the OM and/or IM (b and d respectively). The diffusion through

the periplasm which is packed with proteins is represented by (c). Drugs can be recognised as foreign

molecules by the efflux pumps (e and f) and can be transported out of the membranes. Several pumps

such as AcrA/B and AcrE/F must recruit the OM barrel protein such as TolC (g) in order to expel drugs

directly out of the cell (35). ..................................................................................................................... 7

Figure 1.5: Timeline of antibiotics from their market introduction (red line) to the first observations of

drug resistance, which is represented by a blue ‘X’. The figure is reproduced and adapted for the needs

of this report from Kennedy et al. (37). .................................................................................................. 8

Figure 1.6: Mechanisms through which drug resistance can be spread horizontally from one bacterium to another reproduced from Levy et al. (9). .......................................................................................... 10

Figure 1.7: Mechanism of pro-drug platform, reproduced by Kim Lewis (70). The prodrug enters the cell

and is converted into the reactive drug by a bacteria-specific enzyme (E). Inside the cell, the reactive

compound covalently binds to unrelated targets, T1, T2 and T3, which causes cell death. This mechanism kills not only dividing cell but also dormant cell. ............................................................... 13

Figure 1.8: Native cyclodextrin structures for α-, β- and γ-cyclodextrin (CD). ..................................... 14

Figure 1.9: Chemical structure of rifampicin, gentamicin and chlorhexidine. ....................................... 20

Figure 1.10: Structure of Ra LPS derived from E. coli EH100, Rc LPS derived from E. coli J5 and Re

LPS derived from S. minnesota. .......................................................................................................... 22

Figure 1.11: Langmuir trough lipid monolayers and their transition stages upon compression as described by changes in the Surface Pressure-Area isotherm. The illustration depicts the packing or

Table of Figures

ix

the lipid alcyl chain upon compression from its 2D gas state until the collapse of the monolayer and the

formation of 3D structures. ................................................................................................................... 23

Figure 1.12: The classic optics of a beam, hitting a planar surface, related to the geometry of this process with the wavevector transfer. The wavevector is the “spatial” frequency of the neutron wave

and is described by the 𝑘 = 2𝜋𝜆 , where k is the wavevector and λ is the neutron wavelength. 𝑘𝑖, 𝑘𝑓

and 𝑘𝑟 are the wavevectors of incident, reflected and refracted neutron beam, respectively. 𝑄 is the

wavevector transfer (𝑘𝑖 − 𝑘𝑓), 𝜃𝑖𝑛, 𝜃𝑜𝑢𝑡 and 𝜃𝑟 are the incident, outgoing and refracted angle of the

neutron beam. ...................................................................................................................................... 24

Figure 1.13: (A) The oscillations in the reflectivity profile resulting from the interference which depends on the thickness (d) of the layer (in terms of the position) and the difference in scattering contrasts

between the respective interfaces (in terms of the amplitude). 𝑘𝑡 is the wavevectors of the incident

neutron beam and 𝑘𝑓1 and 𝑘𝑓2 the outgoing neutron beam from the top and bottom of the interfacial

layer, respectively. 𝑛1 and 𝑛2 are the refractive index of light while SLD1 and SLD2 are the refractive

indexes of neutrons in two different bulk phases. 𝜃𝑖𝑛, 𝜃𝑜 and 𝜃𝑟 are the incident, outgoing and refracted

angle of the neutron beam. (B) The scattering geometry in both Cartesian and spherical polar

coordinates with ϕ and θ are angles determining the direction of the outgoing neutron beam. 𝑘𝑓 is the

wavevectors of reflected neutron beam and λ is the neutron wavelength. .......................................... 25

Figure 1.14: Asymmetric lipid bilayer formed of Ra-EH100 LPS and DPPC and deposited on a silicon

surface. The red arrows show the neutron path when the bean hits the surface. ............................... 27

Figure 2.1: Mechanism of tricycle PBD core binding to the N2 of guanine in the DNA minor groove reproduced from Brucoli et al. (143). ................................................................................................... 29

Figure 2.2: Chemical structure of PPA148. .......................................................................................... 30

Figure 2.3: Different representations of the structure and conformation of β-cyclodextrin, showing the

orientation of the hydrogens and hydroxyl groups. Native βCD contains hydroxyl groups, while its

derivatives can be substituted by different functional groups. RAMEB contains either hydrogens or

methyl groups, while HPβCD has either hydrogens or hydroxypropyl groups. .................................... 31

Figure 2.4: Types of phase solubility diagrams (“A” and “B”) and their deviations (“AP”, “AN”, “BS” and “BI”) (154). ............................................................................................................................................ 42

Figure 2.5: 1NMR of sequential reaction monitoring for the synthesis of the final PBD core compound

(#9). ...................................................................................................................................................... 45

Figure 2.6: 1NMR of sequential reaction monitoring for the synthesis of the final tail compound (#11).

............................................................................................................................................................. 46

Table of Figures

x

Figure 2.7: 1NMR signals for the secondary amide and double bond hydrogens of the final de-protected

compound (#12), PPA148. ................................................................................................................... 47

Figure 2.8: 1H-NMR signals for the hydrocarbon chain, methyl moieties, thiomorpholine and the pyrrole group of the PBD tricyclic core of the final compound, PPA148. ......................................................... 47

Figure 2.9: FTIR monitoring of the core synthesis. .............................................................................. 48

Figure 2.10: FTIR monitoring the tail synthesis. .................................................................................. 49

Figure 2.11: FTIR monitoring of the final protected and de-protected PPA148. ................................. 49

Figure 2.12: A representative adsorption profile of PPA148 at the air/liquid interface at 23 °C. The drug was injected below a 0.9% NaCl subphase and changes in surface pressure were recorded until a

plateau was reached. The representative curve was fitted (red line) based on a 3-parameter Hill

equation (Equation 2.4), R2=0.99 (P≤0.0001). The mean and std of maximum surface pressure (12.34

± 0.77 mN/m), Hill slope (1.03 ± 0.22) and t50% (0.33 ± 0.15 h) was calculated from a set of 3 runs. .. 51

Figure 2.13: Representative UV/Vis spectra of PPA148 in three solvents: EtOH, H2O, EtOH/H2O (80:10) at 25 °C. ............................................................................................................................................... 52

Figure 2.14: (A) Effect of solvent and drug concentration on the absorbance profile. Inset: concentration

range where the Beer-Lambert law is followed (𝑦 = 0.04𝑥 + 0.02, 𝑅2 = 0.98 for EtOH/H2O, 𝑦 = 0.04𝑥 +

0.06, 𝑅2 = 0.75 for EtOH, 𝑦 = 0.03𝑥 + 0.05, 𝑅2 = 0.98 for H2O). First derivative of UV/Vis spectrum of a

series of PPA148 concentrations in: (B) H2O; (C) EtOH; and (D) EtOH:H2O (80:20). The peaks and

troughs of the spectrum profile is presented with zero first derivative. The insets show the changes of

λmax with increasing drug concentration. ............................................................................................ 53

Figure 2.15: Increasing concentration of PPA148 in water 7 days under mild stirring at 25 °C. Visual observation of the precipitation point is at 50 μg/mL. ........................................................................... 54

Figure 2.16: Turbidimetric assay: Effect of cyclodextrin concentration in the aggregation of PPA148

using dynamic light scattering (A) and UV/Vis spectroscopy (B). ........................................................ 54

Figure 2.17: Changes in the fluorescence intensity of PPA148 with increasing cyclodextrin

concentration (HPβCD or RAMEB) at 25 ˚C (A). The lines represent the non-linear regression fit

(Equation 2.7) of the experimental data points assuming a 1:1 complex. The phase solubility diagram

(B) was built by adding an excess of drug to solutions of cyclodextrin to measure drug solubilisation. In

both experiments, the solution used was HEPES buffered saline including CaCl2 at pH 7.2. ............ 55

Figure 2.18: Changes in the fluorescence intensity of rifampicin in increasing cyclodextrin concentration

(A) show the drug’s affinity to HPβCD and RAMEB at 25 ˚C. The lines represent the non-linear

regression fit of the experimental data points assuming a 1:1 complex. The phase solubility diagram (B)

was built to observe the approximate amount of drug that can be solubilized with CDs and gave an

Table of Figures

xi

indication of the complex stoichiometry and affinity. In both experiments, the solution used was Tris

buffer pH 9. .......................................................................................................................................... 56

Figure 2.19: Job’s plot for rifampicin/DIMEB (A) and PPA148/DIMEB (B) complexes. In the case of rifampicin, the normalized chemical shift (Δδ*XCD) of the H1 protons of DIMEB were plotted against the

molar fraction of DIMEB (XCD). For, PPA148, protons at position 5 of the cyclodextrin structure. ....... 58

Figure 2.20: Schematic representation of rifampicin/CD inclusion complex as it was found by molecular simulation by He et al. and Tewes et al. (162,164). ............................................................................. 61

Figure 3.1: Structure DPPC (16:0) and DMPG (14:0). ......................................................................... 65

Figure 3.2: Empty Fluidosomes (SUVs, 1mg/mL) were manufactured with the thin film hydration method and probe sonication for size reduction. (A) Effect of liposome integrity in terms of size and intensity

distribution profile of liposomes at a concentration of 1 mg/mL and temperature of 25 ˚C, measured by

dynamic light scattering, over a period of 2 months. The samples were stored at 4 ˚C and tested

between measurement intervals. All samples presented a polydispersity index of approximately 0.2 ±

0.01. (B) Effect of PBS buffer pH 7.4 on the size of a series diluted SUVs (1, 0.5, 0.25, 0.125 and 0.0625

mg/mL) over time. ................................................................................................................................ 73

Figure 3.3: Fluidosome (1 mg/mL SUVs) encapsulated with a series of HPβCD solutions in PBS buffer

pH 7.4 (0.1, 1, 2.5 and 5% w/w) manufactured with thin film hydration method and probe sonication for

size reduction. (A) Effect of HPβCD on the size and polydispersity of Fluidosomes in terms of the intensity distribution profile of liposomes at 25 ˚C, measured by dynamic light scattering. All samples

presented a polydispersity index of approximately 0.3 ± 0.01. (B) Comparison of size changes in the

absence and presence of HPβCD. ...................................................................................................... 74

Figure 3.4: Method development for the separation of unentrapped material from fluidosomes. Empty liposomes (2.5 mg/mL), prepared with thin film hydration method and extrusion through a 100 nm

polycarbonate membrane for size reduction, were passed through a Sephadex PD10G25 size exclusion

chromatography column, washed 8 times and the eluted volumes (8 fractions) were collected and

analyzed. (A) Elution pattern of empty fluidosomes (2.5 mg/mL). The derived count rate profile and lipid

content in each aliquot were measured to investigate the elution of liposomes. (B) Representative

intensity distribution of empty fluidosomes in HEPES buffered saline at pH 7.2 (aliquot #3). ............. 75

Figure 3.5: (A) Elution pattern of PPA148/RAMEB encapsulated fluidosomes (2.5 mg/mL). The derived count rate profile and lipid content in each aliquot were measured to investigate the elution profile of

loaded liposomes. (B). Representative intensity distribution of drug/RAMEB encapsulated fluidosomes

in HEPES buffered saline at pH 7.2 (aliquot #3). ................................................................................. 76

Figure 3.6: (A) Drug quantification profile of the eluted fraction from the PD10G25 size exclusion chromatography column with the inset being the calibration curve (y=0.7937+5.717,R^2=0.9966) is

Table of Figures

xii

presented as a function of the area under the curve (AUC). (B) The UV spectra of pure and extracted

drug in ethanol/water (80:20). Version 1 and 2 and the two UV spectra of extracted PPA148 from aliquot

3, which present changes in drug peak shape. .................................................................................... 77

Figure 3.7: Kirby-Bauer assay for measuring the growth inhibition of Escherichia coli DH5α. The

inhibition zone diameter was measured for PPA148 as a pure substance and as a formulated drug (in

a 5% RAMEB complex and incorporated into 0.1 mg fluidosomes as a complex with 5% RAMEB) after

24 hour incubation at 37 °C. Rifampicin and vancomycin were used as positive and negative control

samples. The asterisks denote the level of significance going from lower to higher (as follows) and four asterisks indicate the highest level: 0.1244 (ns), 0.0332 (*), 0.0021 (**), 0.0002 (***), <0.0001 (****). All

samples were tested in triplicate apart from the final formulation (PPA148-in-RAMEB-in-fluidosomes)

which was tested 5 times. .................................................................................................................... 78

Figure 3.8: Tukey’s test of experimental data (n=3,5) with confidence interval of 95%. (A) Zero is

included in the interval of PPA148-in-CD and PPA148-in-CD-in-liposome which indicates that their mean is not statistically significant. Residuals distribution of experimental observations for each group

(B). ....................................................................................................................................................... 79

Figure 3.9: Monte-Carlo simulation was carried out based on the standard deviation of the experimental

data. 100 observations were generated by the simulation and one-way ANOVA and Tukey’s test was

carried out with 95% confidence interval (A). All sets of groups have statistical difference. Residual distributions of simulated data (B). ....................................................................................................... 79

Figure 3.10: Schematic representation of the possible organization of PPA148/RAMEB complex

incorporated into Fluidosomes. The electrophilic center of PPA148 is presented inside the CD cavity

while the headgroup of DPPC is associated with the exterior hydrogens of the narrow side of RAMEB

molecules. ............................................................................................................................................ 84

Figure 4.1: Langmuir-Blodget and Langmuir-Schafer deposition on a silicon crystal to manufacture the

assymetric model outer membrane of Gram negative bacteria. .......................................................... 95

Figure 4.2: A, C and E present the Surface Pressure-Area (P-A) isotherms generated at 23 °C for a series of pure or mixtures of phospholipids (E. coli extract, DPPC, DPPG, POPC, POPG). All

phospholipids were suspended in chloroform apart from d62DPPC which was prepared in

chloroform:methanol:water (6:4:1) before deposition. The subphase used was 1mM MgCl2 for all

monolayers apart from d62DPPC which was 5 mM CaCl2. The E. coli B monolayer was deposited either

on ultra pure water or 0.9% NaCl The subphase for hydrogenous DPPC/DPPG and POPC/POP

mixtures monolayers was MgCl2. E. coli isotherm was measured on ultra pure water and isotonic saline.

The red arrows show the L1/L2 and the L2/S transition of the lipids where applicable. B, D and F show the calculated compressibility modulus derived from P-A isotherms at 23 °C. The red arrows show

transition phases of the lipids which are more pronounced than in the P-A isotherm. ........................ 99

Table of Figures

xiii

Figure 4.3: Collapse surface pressure (A), area per molecule (B) and compressibility modulus (C) at 30

mN/m of all the phospholipid lipid monolayers ................................................................................... 100

Figure 4.4: (A) Surface Pressure-Area (P-A) isotherms generated at 23 °C for a series of glycolipids: Ra EH100 LPS, Rc J5 LPS and Lipid A. The subphase for Rc-LPS and Lipid A monolayers was MgCl2

while CaCl2 was used for Ra-LPS. The red arrows show the L1/L2 and the L2/S transition of the lipids

where applicable. (B) Compressibility modulus derived from P-A isotherms at 23 °C. The red arrows

show transition phases of the lipids which are more pronounced than in the P-A isotherm. ............. 101

Figure 4.5: Collapse surface pressure (A), area per molecule (B) and compressibility modulus (C) at 30 mN/m of three types of glycolipid monolayers. The Ra-LPS did not reach its collapse pressure and it is

excluded from the figure. ................................................................................................................... 102

Figure 4.6: Representative surface pressure – time isotherms at 23°C of chlorhexidine digluconate (A), rifampicin (B), gentamicin (B), PPA148 (C) and DMSO (0.5, 1 and 1.5 % v/v) (D) against E. coli B

monolayer using 0.9% NaCl as a subphase. The black lines are the fitted curves based on the

mathematical model, Hill curve, when it was applicable. ................................................................... 104

Figure 4.7: CF release from liposomes, manufactured from natural E. coli lipid extract, after administration of PPA148, chlorhexidine digluconate and gentamicin at 25°C. ................................ 105

Figure 4.8: Representative surface pressure – time isotherms at 23°C of rifampicin (A), PPA148 (B),

against DPPC/DPPG monolayer using 1 mM MgCl2 as a subphase. The black lines are the fitted curves

based on the mathematical model when it was applicable. ............................................................... 106

Figure 4.9: Representative surface pressure – time isotherms at 23°C of HPβCD (A) and RAMEB (B), against DPPC/DPPG (3:1) and POPC/POPG (3:1) monolayers using MgCl2 as a subphase. ......... 108

Figure 4.10: Interaction isotherm of 20 μg/mL PPA148 (A), 20 μg/mL rifampicin (B) and 0.1 mg/mL

fluidosomes (C) against different types of OM lipid monolayers (Lipid A and Rc-LPS J5) at constant surface area at 23 °C. The profile was fitted, when applicable, to specific binding Hill slope model to

investigate the kinetics and affinity of the interaction. The black lines are the fitted curves based on the

mathematical model. .......................................................................................................................... 109

Figure 4.11: Pressure-time interaction isotherm of HPβCD (A) and RAMEB (B) with 2 monolayers under

constant surface area at 23°C. .......................................................................................................... 110

Figure 4.12: Neutron reflectivity profile (A) and model data fits with their scattering length density profiles

(B) for an assymetrically deposited DPPC (inner leaflet) and Ra-LPS (outer leaflet) model membrane

at room temperature. The sample was measured at three isotopic contrasts (100% D2O, 100% H2O

and SMW). The model membrane was fitted into five-layered mathematical model: Silicon oxide (SiO2),

DPPC headgroup (Inner HG), DPPC tails (Inner tails), Ra LPS tails (Outer tails) and Ra LPS headgroups (Core). ............................................................................................................................ 111

Table of Figures

xiv

Figure 4.13: Neutron reflectivity profile (A and model data fits with their scattering length density profiles

(B) for an assymetrically deposited DPPC (inner leaflet) and Ra-LPS (outer leaflet) model membrane

at 38 °C. The sample was measured in three isotopic contrasts (100% D2O, 100% H2O and 38% D2O/SMW). The model membrane was fitted into a five-layered mathematical model: Silicon oxide

(SiO2), DPPC headgroup (Inner HG), DPPC tails (Inner tails), Ra LPS tails (Outer tails) and Ra LPS

headgroups (Core). ............................................................................................................................ 113

Figure 4.14: Neutron reflectivity profile (A) and model data fits with their scattering length density profiles

(B) for an assymetrically deposited DPPC (inner leaflet) and Ra-LPS (outer leaflet) model membrane after being challenged with 0.1 mg/mL fluidosomes (DPPC/DMPG, 18:1) at 38 °C. The sample was

measured in three isotopic contrasts (100% D2O, 100% H2O and 38% D2O/SMW). The model

membrane was fitted into a seven-layered mathematical model: Silicon oxide (SiO2), DPPC headgroup

(Inner HG), DPPC tails (Inner tails), Ra LPS tails (Outer tails) and Ra LPS headgroups (Core), Bridge

and Floating bilayer. ........................................................................................................................... 114

Figure 4.15: Representation of the possible structural changes in the model OM system after being challenged by fluidosomes. ................................................................................................................ 116

Figure 4.16: Schematic representation of the movement of phospholipids within a membrane: transpose

diffusion (1), rotation (2), swing (3), flexion (4) and transverse diffusion or “flip-flop” movement (5). 118

Figure 4.17: Schematic representation of the effect of temperature change on membrane structure and behavior of lipid bilayers adapted from Los and Murata (249). Low temperatures cause “rigidification”

of membranes, whereas high temperatures cause “fluidization” of membranes. .............................. 119

Figure B 1: 13CNMR of intermediate compounds of the PBD core synthesis. The samples were

suspended in deuterated chloroform. ................................................................................................ 145

Figure B 2: 13CNMR of intermediate compounds of the tail synthesis. The samples were suspended in deuterated DMSO. ............................................................................................................................. 145

Figure B 3: 13CNMR of the final compound, PPA148. The sample was suspended in deuterated

chloroform. ......................................................................................................................................... 146

Figure B 4: The linear regression (A) met the assumptions that the residuals follow a normal distribution (B) and their variance is constant (B). ................................................................................................ 146

Table of Tables

xv

Table of Tables Table 1.1: Comparison of non-antibiotic drug properties with the generic optimal properties for

antibacterial drugs for oral or parental administration as were calculated by O’Shea et al based on data

collected from the literature and drug data bases (12). ......................................................................... 4

Table 1.2: WHO priority list for the development and discovery of new antibiotics for antibiotic resistant bacteria (54). ........................................................................................................................................ 11

Table 1.3: Summary of antibiotic-cyclodextrin complexes investigated by research groups to achieve

sustained drug release, better bioavailability, increased antimicrobial activity and enhanced drug

solubility. .............................................................................................................................................. 16

Table 1.4: Several liposomal carriers investigated against Gram negative bacteria. .......................... 17

Table 1.5: Physicochemical properties of PPA148, rifampicin, gentamicin and chlorhexidine. ........... 20

Table 1.6: Fatty acid composition of E. coli lipid extract as it was found by White et al. (127). ........... 21

Table 1.7: Potential isotopic contrasts and main information content of a monolayer on an air/water

interface or a deposited lipid bilayer on a silicon surface consisting for tail deuterated phospholipids.

NRW and SMW stand for null reflective water (8.2% D2O) and silicon match water (38% D2O) respectively and are used to cancel out the background of the two mediums and highlight the interface.

............................................................................................................................................................. 26

Table 2.1: Comparison of the physicochemical properties of PPA148 estimated by ChemDraw

compared (the experimental values are presented in parenthesis) with the generic optimal properties

of antibacterial drugs for oral or parental administration as calculated by O’Shea et al. (12) based on data collected from the literature and drug data bases. ....................................................................... 30

Table 2.2: Chemical characteristics of each intermediate and the final compound. ............................ 50

Table 2.3 Solubility values and aggregation concentration of PPA148 in water, HEPES buffered saline pH 7.2 and HPβCD or RAMEB in HEPES obtained by different techniques. ...................................... 54

Table 2.4: Binding constant of PPA148 and rifampicin as obtained by two different techniques. ....... 57

Table 2.5: Gibbs free energy of transfer of rifampicin and PPA148 from pure water to aqueous CD solutions at room temperature calculated using Equation 2.11. .......................................................... 57

Table 4.1: Summary of scattering length densities of the lipid components studied, and the solution

subphases. ........................................................................................................................................... 96

Table of Tables

xvi

Table 4.2: Kinetic parameters obtained from the fitting of the binding isotherms of rifampicin, gentamicin

and PPA148 and the E. coli B monolayers at 23°C on water sub-phase containing isotonic saline. 104

Table 4.3: Kinetic parameters obtained from the fitting of the binding isotherms of rifampicin and PPA148 and the DPPC/DPPG monolayers at 23°C on water sub-phase containing1 mM MgCl2. ... 107

Table 4.4: Kinetic parameters obtained from fitting the binding isotherms of rifampicin and PPA148 and the Rc J5 LPS and R595 Lipid A monolayers at 23°C on a water sub-phase containing1 mM MgCl2.

........................................................................................................................................................... 109

Table 4.5: Volume fractions of deuterated DPPC tails, hydrogenous LPS tails and water within the

bilayers of Ra-LPS/d-DPPC at room temperature (25 °C). ................................................................ 111

Table 4.6: Structural parameters, thickness, hydration and roughness, of the Ra-LPS/DPPC membrane at room temperature and 38 °C as obtained from the fitting of the neutron reflectivity data. The numbers

in parentheses are the 95% confidence interval error. ...................................................................... 112

Table 4.7: Volume fraction of deuterated DPPC tails, hydrogenous LPS tails and water within the bilayers of Ra-LPS/d-DPPC 38 °C before and after the challenge by the fluidosomes. .................... 113

Table 4.8: Structural parameters, namely, thickness, hydration and roughness, of the challenged Ra-

LPS/DPPC membrane at 38 °C as obtained from fits to the neutron reflectivity data shown in Figure

4.14. ................................................................................................................................................... 115

Table B 1: Minimum Inhibitory concentrations (MIC) of PPA148 against Gram-negative bacteria in the

absence and presence of the efflux pump inhibitor PaβN. ................................................................ 144

Table B 2: Minimum Inhibitory concentrations (MIC) of PPA148 against Gram-positive bacteria in the

absence and presence of the efflux pump inhibitor PaβN. ................................................................ 144

Table C 1: Number and Volume distribution of empty and loaded fluidosomes. ............................... 147

Abbreviations

xvii

Abbreviations ABC: ATP binding cassettes

AMR: Antimicrobial Resistance

CE: complexation efficiency

CL: cardiolipins

CymA: cyclodextrin metabolism A

DCR: derived count rate

DIMEB: Heptakis (2,6-di-O-methyl)-β-cyclodextrin

DNA: deoxyribonucleic acid

EARS-Net: European Antimicrobial Resistance Surveillance Network

ECDC: European Center of Disease Control

IM: Inner membrane

IUPAC: International Union of Pure and Applied Chemistry

Kdo: 3-deoxy-d-manno-oct-ulosonic acid

LC-MS: Liquid Chromatography-Mass Spectrometry

LPS: Lipopolysaccharide

MATE: multidrug and toxic-compound efflux

MDR: multi-drug resistant

MIC: minimum inhibitory concentration

MW: molecular weight

NRW: null reflected water

OM: Outer mambrane

Abbreviations

xviii

OMP: Outer membrane protein

PC: phosphatidylcholine

PE: phosphatidylethanolamines

PG: phosphatidylglycerols

PSA: polar surface area

RNA: ribonucleic acid

RND: resistance-nodulation-division

SMR: small multidrug resistance

SMW: silicon matched water

US CDC: US Center for Disease Control and Prevention

WHO: World Health Organization

Symbols

xix

Symbols (m:n): stoichiometry of CD/drug complex

ΔGtro: Gibbs free energy

A: area per molecule

Cs: Area compression modulus

D: Brownian motion diffusion coefficient

d: thickness

dH: hydrodynamic diameter

Es or Cs-1: Compressibility modulus

f: wave frequency

G: gaseous state

h: Hill slope

I: Fluorescence intensity

K: binding constant

k: wavevector

Ka: distance travelled by the analytefrom its origin

Ks: distance travelled by the solvent from its origin

L1: liquid-expanded state

L2: condensed state

p: momentum of neutrons

Q: wavevector transfer

Rf: retardation factor

Symbols

xx

S: solid state

S0: Solubility in the presence of cyclodextrin

Sint: intrinsic solubility

T: Temperature in Kelvin

v: neutron velocity

XCD: mole fraction of cyclodextrin

ΔΠ: Difference in surface pressure

ΔΠ/ΔA: slope of isotherm

η: solvent viscosity

θ: incident angle

λ: wavelength

Π: Surface pressure

ρ: Scattering Length Density (SLD)

φ: volume fraction

Physical Constants

xxi

Physical Constants h: Planck's constant, 6.626 × 10−34 J

k: Boltzman’s constant, 1.381 × 10−23 J K-1

mn: neutron mass, 1.675 × 10−24 kg

R: Gas constant 8.314 JK-1mol-1

General Introduction

1

Chapter 1 General Introduction


2

1.1 Antibiotics and Bacteria: a unique class of medicines and

pathogens

Antibiotics are a group of drugs used to treat bacterial infections. They are a unique class of

medicines because of their complicated mechanisms of action as well as their restricting use and long-term efficacy. Unlike other drugs, antibiotics should selectively target several pathogens

that are responsible for conditions ranging from pneumonia to urinary infections, acting in

different body sites, whilst at the same time they should not affect the host organism (1).

Antibiotics made a significant contribution to public health, having been used not only in medical

practice but also in agriculture and other industries. They also help to promote growth in livestock,

preserve building materials from contamination and treat blight in orchards (2). Antibiotics have

saved countless lives and improved surgery conditions by preventing or curing bacterial

infections in patients including those: receiving chemotherapy treatments; suffering from chronic diseases such as cystic fibrosis, diabetes, end-stage renal disease, or rheumatoid arthritis; or

having undertaken complex surgical procedures such as organ transplants, joint replacements,

or cardiac surgery (3–5). Moreover, they have increased the average life expectancy and

decreased the morbidity and mortality caused by food-borne and other poverty-related infections

(5,6).

1.2 Antibiotic classification

Antibiotics are classified based on their chemical family and mechanism of action. They are

separated into 32 groups, based on their chemical functional groups, the larger being aminoglycosides, glycopeptides, imidazole, macrolides, nucleosides, penicillins and

cephalosporines (β-lactams), peptides, pyrimidines and pyridines, sulfonamides and

tetracyclines. Each of them possesses unique chemical properties and functional groups to target

different parts of bacterial cells. For example, penicillins inhibit cell wall synthesis (7), whilst

rifampins target RNA synthesis (8). Their action to either kill bacteria (bacteriocidal) or inhibit

bacterial growth (bacteriostatic) falls within four categories. Three of those functions are the

inhibition of (a) enzymes involved in cell wall and protein synthesis, (b) deoxyribonucleic acid (DNA) replication and/or (c) ribonucleic acid (RNA) transcription as shown in Figure 1.1 (9,10).

The fourth category is the disruption of bacterial membrane structure (10). For example,

penicillins; cephalosporins; carbapenems; daptomycin and monobactams; glycopeptides cause

inhibition of cell wall synthesis. Tetracyclines, aminoglycosides, oxazolidonones, streptogramins,

ketolides, macrolides and lincosamides cause inhibition of protein synthesis. Fluoroquinolones

act by inhibiting DNA synthesis while rifampins RNA synthesis. Sulfonamides cause inhibition of

folate synthesis. Daptomycin depolarizes membrane potential (10).


3

Figure 1.1: Mechanism of action of some major antibiotic agents reproduced from Clatworthy et al. (10).

1.3 Bacterial classification

Bacteria are prokaryotic microorganisms which can be classified as either Gram positive or Gram negative, on the basis of their cell envelope structure. The cell envelope of Gram positive bacteria

possesses a thick peptidoglycan wall external to their plasma membrane, whereas the

peptidoglycan of Gram negative constitutes a thinner layer between their outer and inner

membranes.

The activity of antibiotics is designated as broad, intermediate or narrow-spectrum based on the number of different bacterial species they can affect (11). For example, antibiotics that kill both

Gram positive and negative bacteria are called broad spectrum antibiotics, while those which kill

only Gram positive bacteria are considered narrow spectrum (11). The research conducted in

this project is focused exclusively on Gram negative bacteria because their complex outer

membrane (OM) hinders drug uptake, resulting in low efficacy.

1.3.1 Physicochemical properties of antibiotics

Active pharmaceutical compounds are classified based on their physicochemical properties and

fall within the Lipinski's rule of five (i.e., a molecule with a molecular mass under 500 Da, no more

than 5 hydrogen bond donors, no more than 10 hydrogen bond acceptors, and an octanol–water

partition coefficient, logP, no greater than 5). Antibacterial agents have always been deemed to

deviate from Lipinski’s rule of five due to their higher molecular weight and polarity (Table 1.1)

(12,13). Several research groups have tried to correlate the physicochemical properties of

antibiotics with their activity by collecting experimental data from the literature and databases of already marketed drugs to compare them in terms of lipophilicity (clogP), polarity (PSA),


4

molecular weight (MW), number of H-donors, number of H-acceptors and number of aromatic

rings. Leeson and Davis compared drugs, including antibiotics, from five different therapeutic

areas (cardiovascular, nervous system, gastrointestinal, respiratory/inflammation and infection) that were launched from 1983 to 2002 (14). Antibiotics, antimalarial, antiviral, antifungal and

antiparasitic drugs were all included in the general infection category, known as anti-infectives.

Their results showed that anti-infectives have a different physicochemical profile by being larger

and less lipophilic than the other classes, which can be explained by their activity against non-

human cells and their need to penetrate the bacterial cell envelope. O’Shea and co-workers,

particularly, examined the properties of 147 antibacterial agents, classifying the results into

activity against Gram negative and Gram positive bacteria, and compared them with results from non-antibiotic drugs (12). Their analysis is presented in Table 1.1.

Table 1.1: Comparison of non-antibiotic drug properties with the generic optimal properties for antibacterial drugs for oral or parental administration as were calculated by O’Shea et al based on data collected from the literature and drug data bases (12).

Physicochemical properties Non-antibiotic

Antibiotics with optimal properties against Gram

negative

Antibiotics with optimal properties

against Gram positive

Molecular weight (g/mol)

338 414 813

cLogP 2.7 -0.1 2.1

cLogD7.4 1.6 -2.8 -0.2

PSA (Å2) 70 165 243 cLogP is the logarithm of the partition coefficient (LogP) between n-octanol and water and is a measure of the compound’s hydrophilicity and an indication of absorption and permeation. clogD7.4 is the logarithm of the distribution coefficient at pH 7.4 which is another measure of lipophilicity of unionized and ionized forms. PSA is the polar surface area, which provides a prediction of drug passive intestinal permeability (<140 Å2) and blood-brain barrier penetration (<60 Å2) (15–17).

Table 1.1 shows that in general, antibiotics have significantly higher molecular weights and

polarities compared to non-antibiotic drugs. Interestingly, there are variations between drugs

acting against Gram negative and positive bacteria which can be explained by the differences in

the constitution of the cell envelopes between these two groups. The lipophilicity of non-

antibacterial drugs and the Gram positive activity group present similarities, but a substantially

higher polarity is noted for the Gram negative specific group. This difference in lipophilicity is

more obvious when comparing the cLogD7.4 (4 log units lower, which means more hydrophilic)

and PSA values.

1.3.2 Gram negative bacterial cell envelope as a barrier to antibiotics

Gram negative bacteria possess a complex envelope structure with two main permeability

barriers: the outer membrane (OM) and the inner (cytoplasmic) membrane (IM). The OM is a

unique feature of the Gram negative bacterial cell envelope providing cell protection. It is a lipid


5

bilayer whose outer leaflet consists of lipopolysaccharide (LPS) while the inner leaflet is

composed of phospholipids (18).

Figure 1.2: Cell envelope of Gram-negative bacteria reproduced from Silhavy et al. (18).

LPS belongs to the category of glycolipids and consists of Lipid A, a hydrophilic oligosaccharide

core and a hydrophilic O-antigenic polysaccharide side chain (O-chain) (Figure 1.2). Lipid A

provides the membrane anchor for the LPS molecule, and is made up of a β(1 → 6)-linked

glucosamine disaccharide backbone which is acylated at positions 2 and 3 of each monosaccharide portion involving six to seven fatty acid chains (Figure 1.10). This acylated

disaccharide is also mostly phosphorylated at positions 1 and 4′ or even further substituted (19–

21). The core oligosaccharide is subdivided into an inner and outer core. The inner core is

composed of 3-deoxy-d-manno-oct-ulosonic acid (Kdo) and L-glycerol-D-manno-heptose which

are typically phosphorylated or modified with phosphate containing groups, increasing their

overall negative charge helping stabilize the structure through divalent cation cross-linking (22).

Kdo links the core oligosaccharide with Lipid A anchor while the outer core attaches to the O-

antigen. The outer core is more diverse than the inner and is composed of three to six additional saccharides such as D-glucose and N-acetyl-D-glucosamine and D-galactose (23,24). There are

also proteins present in the outer membrane (OMP); lipoproteins and β-barrels. OMP such as

porins are of major importance because they are transmembrane and their function is to allow

the uptake of nutrients and hydrophilic compounds. LPS provides protection against hydrophobic

compounds (18), and makes the OM a semipermeable barrier which protects the cell from

harmful organic compounds such as antimicrobials and allows the uptake of sufficient nutrients.

A thin peptidoglycan layer sits between the OM and IM to provide extra support and flexibility to

the IM, which is the last part of the cell envelope and is composed of phospholipids and proteins.

IM proteins are responsible for the energy production, lipid biosynthesis, protein secretion and

transport of materials in and out of the cell. The principal lipids which are present in the IM of the

majority of bacteria, such as P. aeruginosa and Escherichia coli (E. coli), are


6

phosphatidylcholines (PCs), phosphatidylethanolamines (PEs), phosphatidylglycerols (PGs) and

cardiolipins (CL) (25) (Figure 1.3). Bacterial efflux pumps present in the cell envelope prevent

the accumulation of toxic compounds, such as antimicrobial drugs, in the cell. Efflux pumps are comprised of a transporter (protein) in the IM, an accessory protein (membrane-fusion protein)

in the periplasm and an outer membrane protein channel in the OM (Figure 1.4 e, f, g) (26).

Figure 1.3: Structure of phospholipid headgroups with different substitution in the phosphate group which gives a different curvature and charge. R1 and R2 are the fatty acid chains which can be either saturated (PC with C16:0 is DPPC while PC with C14:0 is DMPC) or unsaturated (PC with C16:1 is POPC).

1.3.3 Drug uptake mechanisms

Based on the cell envelope structure (Figure 1.2) it can be seen that there are several

mechanisms of drug and ion uptake (Figure 1.4). They are divided into two major categories: passive and active transport (27). Hydrophobic compounds diffuse passively through the

membrane because of their lipid solubility (passive diffusion) while hydrophilic agents are

transported through the water channels (facilitated diffusion). General diffusion porins form water

channels facilitating the passage of small hydrophilic molecules (~ 400 Da, >600 pass very

slowly) and ions (non-specific diffusion) while the specific porins allow the passage of specific

molecules or classes of molecules that binds to the substrates that preferentially flow through

these channels (self-promoted diffusion) (18,28–30). For instance, it has been found that uptake of native α- and β-cyclodextrin depends on specific membrane protein channels (cyclodextrin

metabolism A or CymA) in the outer membrane of the Gram negative bacterium, Klebsiella

oxytoca (31). Drugs might have a different mechanism of uptake across OM and IM as is the

case with the aminoglycosides. Gentamicin is known to pass passively (self-promoted) through

the OM and actively through the IM (32,33).

Phosphatidylethanolamine (PE)

Phosphatidylglycerol(PG)

Phosphatidylcholine (PC)

Cardiolipin(CL)


7

Figure 1.4: Different ways by which a substance can be transported across the membranes reproduced from Bolla et al., (34). Hydrophilic components and nutrients can pass through porins (a) and hydrophobic molecules diffuse through the OM and/or IM (b and d respectively). The diffusion through the periplasm which is packed with proteins is represented by (c). Drugs can be recognised as foreign molecules by the efflux pumps (e and f) and can be transported out of the membranes. Several pumps such as AcrA/B and AcrE/F must recruit the OM barrel protein such as TolC (g) in order to expel drugs directly out of the cell (35).

1.4 Antimicrobial Resistance (AMR)

Although antibiotics have been highly effective against pathogens, they impose substantial

selective pressures on bacteria, which are living organisms and thus tend to adapt to changes

and evolve. Antibiotic-resistance is a natural process because bacteria develop their own

resistance mechanisms and constantly acquire those that had evolved over billions of years in

environmental bacteria (36). Therefore, resistance is not a new phenomenon, as it is shown in

Figure 1.5. It first emerged in hospitals in the late 1930s with the sulfonamide-resistant Streptococcus pyogenes bacteria, shortly after sulfonamides were made available on the market,

and penicillin-resistant Staphylococcus aureus appeared 6 months after being used in clinical

practice in the UK (36–38). After 1985, the antibiotic pipeline began to dry up and fewer new

drugs were introduced; leading to a situation now, where we are in danger of entering a “post-

antibiotic” era (2,10,36,37,39–41).


8

Figure 1.5: Timeline of antibiotics from their market introduction (red line) to the first observations of drug resistance, which is represented by a blue ‘X’. The figure is reproduced and adapted for the needs of this report from Kennedy et al. (37).

1.4.1 Causes of AMR

Antimicrobial resistance is driven by natural factors. The results of human activity, clinical use,

extensive use of antibiotics in agriculture and public behavior and perception have all contributed

to its widespread proliferation. Bacteria are small prokaryotic organisms with simple intracellular

features and circular DNA (chromosomal DNA) which coexists with independent DNA pieces

(plasmids). Plasmids encode traits and genes required for ever-varying environmental

challenges. Due to the small size of bacterial cells, individual pathogens contain a limited number

of those genes, thus hindering its ability to face environmental changes. The key to bacterial

natural adaptive strength is their ability to reproduce rapidly, creating bacterial populations with wide variability in terms of advantageous characteristics. Even though changes in the genes of

plasmids are rare, the rapid generation time ensures that these advantageous changes will

become rapidly predominant in the bacterial population and help them evolve in response to the

environmental changes.

Although antimicrobial resistance is a natural process, the main reason for the widespread proliferation was both the misuse and overuse of antibiotics in humans and animals. Once

antibiotics were introduced, they were effective and efficient. Thus, there was a general public

belief that antibiotics should be applied in the first instance for every type of illness. This

perception, along with the over-prescription and extended regimens in clinical practice, created

strong selective pressures on bacteria which, in turn, generated resistant strains responding to

prolonged antibiotic exposure. The more bacteria are exposed to antibiotics, the more likely the

emergence of resistant strains. Empirical treatment and prescription of antibiotics by general

practitioners based on local epidemiology before the results of the diagnostic tests often leads to their misuse. If the initial antibiotic regimen is not the appropriate one, the patient’s microbiota is

subjected to an intense and repeated selective pressure that encourages and conserves the

development of AMR among currently non-pathogenic organisms. In addition, poor hygiene and

infection prevention leads to an increased number of infections and thus increased use of


9

antibiotics. Moreover, the use of antibiotics in agriculture too, may have increased productivity

but it has been increasing the abundance and diversity of AMR genes across rural and urban

environments.

Bacteria have become resistant to a number of drugs because they have resistance mechanisms

to first and second line of infection treatment. The deficiency of existing antibiotics and the

widespread proliferation of antimicrobial resistance has been examined in depth by the US

Center for Disease Control and Prevention (CDC) (42), European CDC (ECDC) (43), WHO (44)

and European Antimicrobial Resistance Surveillance Network (EARS-Net) (43). These organizations reported that Gram negative bacteria are becoming pan-drug or extensively-drug

resistant which leads to increased morbidity and mortality.

1.4.2 Mechanisms of bacterial resistance

Resistance to antibiotics falls into two broad categories: intrinsic and acquired. The conventional

example of intrinsic resistance is the multi-drug resistant (MDR) phenotype existing in Gram negative bacteria which includes inherent active drug efflux and low permeability of the OM. As

mentioned previously, the Gram negative bacterial outer membrane outer leaflet consists of LPS

which is anchored by Lipid A and stabilized by divalent cation crosslinking between the

phosphate groups which decorate the inner core oligosaccharides (Figure 1.10). This cross-

linking facilitates tight packing of the Lipid A hydrocarbon chains, decreasing OM fluidity and

increasing the permeability threshold for drugs whose main mechanism of uptake is passive

diffusion (45). For those drugs requiring other means of being transported into the bacterial cell

envelope, the limiting step is the structure and properties of the porins and proteins present in the OM (45). Porins selectively retard the influx of drugs and nutrients according to their size,

hydrophobicity and charge. The OM can also control the efflux of materials via efflux pumps

which primarily actively transport toxins out of the cell. Efflux pumps can either be substrate

specific, and only export one molecule, or they can be broad-spectrum, and export structurally

distinct classes of molecules (26). There are five types of efflux pumps: ATP binding cassettes

(ABC), major facilitators (MF), multidrug and toxic-compound efflux (MATE) pumps, small

multidrug resistance (SMR) pumps, and the resistance-nodulation-division family (RND). However, only the latter is the main facilitators of intrinsic resistance (26,46,47). RNDs are also

known as outer membrane proteins and are transmembrane. For example, MexAB-OprM

contributes to β-lactam and fluoroquinolone intrinsic resistance in P. aeruginosa (48) and the

AcrAB-TolC in E. coli facilitates resistance to a broad range of antibiotics such as the

tetracyclines, fluoroquinolones, β-lactams and the macrolides (26). The high level of intrinsic

resistance in Gram negative bacteria is a result of the synergistic effect between the low

permeability of the outer membrane and the efflux pumps. Mutations in genes encoding efflux

pump expression may lead to their upregulation which is clinically associated with the multi-drug resistance phenotype of bacteria (9,26,29,34).


10

Acquired antimicrobial resistance mechanisms includes genetic modifications in the efflux pumps

and porins or incorporation of new genetic material through plasmids, bacteriophages,

transposons and naked DNA (Figure 1.6). The genes for single drug resistance traits developed from a mutation in an intrinsic chromosomal gene can spread from one generation to another

(vertical gene transfer). These genetic mutations can not only cause upregulation of efflux pumps

but also alter antibiotic targets or cause target overexpression (9,45). Moreover, acquired

resistance can occur by transferring genetic material between related or unrelated species

(horizontal gene transfer), which accounts for the rapid proliferation of acquired resistances.

Multiple genes can accumulate in the same organism, leading to multi-, pan- or extensive- drug

resistance (9,45).

Figure 1.6: Mechanisms through which drug resistance can be spread horizontally from one bacterium to another reproduced from Levy et al. (9).

1.5 Strategies to combat antimicrobial resistance

Several reviews have stated that the last resorts for treatment of infections caused by resistant

bacteria were the broad spectrum carbapenem agents. Over recent years, the emergence of

carbapenem-resistant Enterobacteriaceae as well as carbapenem-resistant Gram negative bacilli (Acinetobacter baumannii, Pseudomonas aeruginosa) have left a few treatment options,

while the spread of carbapenem-resistant Enterobacteriaceae has created a serious threat to

public health worldwide (49). In this context, older drugs with adverse effects, such as polymyxins

(colistin and polymyxin B), have been used by clinicians. The re-introduction of polymyxins for

antimicrobial therapy has been followed by an increase in reports of acquired polymyxin

resistance in clinical isolates such as P. aeruginosa, A. baumannii and Klebsiella pneumoniae

(50). Therefore, there is an urgent need to overcome antimicrobial resistance by either discovering new antibiotics, applying combined antibiotic therapy or developing new drug carriers

and rapid diagnostics.


11

1.5.1 Drug development

The development of new drugs was initiated by modifying already marketed antibiotics to extend

their spectrum activity (51). Examples of this initiative are penicillins, fluoroquinolones and cephalosporins. However, bacteria also became resistant to those drugs because the

modifications in their structure failed to change their site and mechanism of action. Attempts have

been made by research groups to introduce new drugs but the majority of these attempts have

failed because their physicochemical properties reduce their bioavailability and reduce outer

membrane penetration. Teixobactin was discovered in 2015 as a new class of antibiotics against

Gram positive bacteria but its activity is limited. Therefore, new drugs need to be discovered as

was proposed by O’Neill (52) in his wide-ranging report into antimicrobial resistance strategies

for the UK government. Table 1.2 presents a list of resistant bacteria deemed by the WHO as urgently requiring the discovery and development of new antibiotics in order to tackle them (53).

1.5.2 Antibiotic combination therapy

The rise of multi-drug resistant (MDR) Gram negative bacteria and the lack of new antibiotic

classes make monotherapy inadequate for treating infections and increases the likelihood that

combination therapy would be more effective (54). Antibiotic combination therapy is widely used to treat severe infections caused by Gram negative bacteria and may be divided into two

categories: antibiotic synergism and antibiotic/adjuvant combination (55,56).

Table 1.2: WHO priority list for the development and discovery of new antibiotics for antibiotic resistant bacteria (57).

Priority 1: Critical Acinetobacter baumannii, carbapenem-resistant

Pseudomonas aeruginosa, carbapenem-resistant Enterobacteriaceae, carbapenem-resistant, 3rd generation cephalosporin-resistant

Priority 2: High Enterococcus faecium, vancomycin-resistant

Staphylococcus aureus, methicillin-resistant, vancomycin intermediate and resistant Helicobacter pylori, clarithromycin-resistant Campylobacter, fluoroquinolone-resistant Salmonella spp., fluoroquinolone-resistant

Neisseria gonorrhoeae, 3rd generation cephalosporin-resistant, fluoroquinolone-resistant

Priority 3: Medium Streptococcus pneumoniae, penicillin-non-susceptible

Haemophilus influenzae, ampicillin-resistant Shigella spp., fluoroquinolone-resistant


12

1.5.2.1 Combination of two or more antibiotics

The combination of antimicrobial agents to produce potent synergistic effects has been applied to combat MDR in Gram negative bacteria. Combination therapy of two to four different antibiotics

against multidrug-resistant Pseudomonas spp. typically includes a broad-spectrum beta-lactam

and an aminoglycoside or a fluoroquinolone, colistin, a macrolide, or rifampin (55,58). The

mechanisms of synergistic bacteriostatic activity are not fully understood for all drug combination

but plausible explanations exist for some antibiotics. For example, colistin increased membrane

permeability by acting as a disruptive agent, enabling rifampin to exert its action against colistin-

resistant Klebsiella pneumoniae carbapenemase (KPC)-producing bacteria (59). Co-administration of colistin and sulfamethoxazole causes an increased in vitro activity against

colistin resistant and susceptible Gram negative bacterial strains when compared to

monotherapy (50). The combination of colistin with meropenem or doripenem has also resulted

in synergistic effects in vitro against multidrug-resistant Pseudomonas spp., Acinetobacter spp.,

and carbapenemase-producing Enterobacteriaceae and has been reported as a successful

treatment in case reports (55).

1.5.2.2 Antibiotic-adjuvant

Antibiotic-adjuvant combination is an alternative approach for reviving antibacterial drug

discovery by targeting the resistance processes (60). Adjuvants are compounds that enhance

antibacterial drug activity but do not act as antibacterial agents themselves. Products consisting

of beta-lactam antibiotics and an inhibitor of the beta-lactamase enzyme are already on the

market and represent a successful approach in targeting resistance to a defined class of antibiotics (61,62). A well-known example is the combination of amoxicillin with clavulanic acid,

which is a beta-lactamase inhibitor but is not effective as an antibiotic by itself (63). There are

newly synthesized beta-lactamase inhibitors for co-administration with cephalosporins and

carbapenemes, i.e. avibactam (64) and vaborbactam (65), respectively. The combination of

avibactam and ceftazidime has recently been approved for therapeutic use in the US and Europe

(66).

Within a similar conceptual framework, efflux pump inhibitors have also been investigated as

adjuvants. A common synthetic efflux pump inhibitor, widely used in vitro, is phenylalanine-

arginine β-naphthylamide (PAβN), capable of inhibiting clinically relevant pumps for resistance

in P. aeruginosa, as well as similar pumps in other MDR Gram-negative bacteria (63,67). The

combination of PAβN and levofloxacin decreased resistance to the drug by eight-fold in wild-type

strains of P. aeruginosa, while in efflux pump overexpressing strains it increased efficacy by up to 64-fold (68). In addition, PAβN has been shown to have membrane permeabilizing activity in

a concentration dependent manner. It promotes its own entry into the cells where it can access

its efflux pump targets (69). Therefore, PAβN has an excellent potential as an antibiotic adjuvant

by increasing outer membrane permeability and/or impairing drug efflux pumps.


13

1.5.2.3 Pro-drug formation

A promising platform to overcome antimicrobial resistance is to design and develop prodrug antibiotics which are ineffective until they are activated by specific bacterial enzymes, allowing

them to covalently bind to unrelated targets (Figure 1.7) (70). The irreversible binding to the target

overcomes the efflux from the MDR pumps, resulting in increased drug penetration and

accumulation into the cells. Prodrug antibiotics were discovered in 1950 in an attempt to eliminate

antibiotics without specific targets, but screening tests eliminated them from being used (70). For

instance, metronidazole, which was overlooked as an antibiotic, is an ideal prodrug with broad

spectrum activity because it becomes non-specific after activation by binding to different targets (proteins and DNA) (71). Fleck and co-workers investigated three new nitrofuran prodrugs

(ADC111, ADC112, ADC113) which all showed sterilizing capability and broad-spectrum activity

(72).

Figure 1.7: Mechanism of pro-drug platform, reproduced by Kim Lewis (70). The prodrug enters the cell and is converted into the reactive drug by a bacteria-specific enzyme (E). Inside the cell, the reactive compound covalently binds to unrelated targets, T1, T2 and T3, which causes cell death. This mechanism kills not only dividing cell but also dormant cell.

1.6 Formulation approaches for antimicrobials to overcome

resistance

1.6.1 Cyclodextrin-drug complexes

Cyclodextrins, also known as cyclomaltodextrins and cycloamyloses, were first identified as a

bacterial digestion product of starch (73,74). Three native cyclodextrins have been isolated: α-

cyclodextrin (αCD), β-cyclodextrin (βCD) and γ-cyclodextrin (γCD). They are cyclic

oligosaccharides containing 6 (αCD), 7 (βCD) or 8 (γCD) (α-1,4)-a-D-glucopyranose units respectively. Due to the chair configuration of the glucopyranose unit, cyclodextrins are toroidal

molecules with a truncated cone structure having a lipophilic inner cavity and hydrophilic outer

surface (Figure 1.8) (75).


14

Figure 1.8: Native cyclodextrin structures for α-, β- and γ-cyclodextrin (CD).

Despite the presence of numerous hydroxyl groups on the exterior of the CD molecule, natural

CDs have limited water solubility. Both αCD and γCD have higher solubility (145 and 232 mg/mL

respectively) than βCD (18 mg/mL at 25 °C) (76). The low water solubility is a result of CD self-

association in an aqueous environments, which is concentration-dependent (73,77,78). CDs form

channel-type aggregates, which are connected through hydrogen bonding between the

cyclodextrin molecules (79). The geometry and axial symmetry of the intermolecular hydrogen bonded aggregates possibly affects their interaction with the surrounding water structure. The 7-

fold symmetry of βCD creates perturbations in the water structure resulting in their abnormally

low water solubility (80).

Advances in biotechnology have led to the synthesis and modification of CDs, enhancing their

water solubility. Substitution of any hydroxyl group (primary and/or secondary) yields a dramatic increase of their water solubility (81). Methylated or hydroxypropyl derivatives of βCD and γCD

are of pharmaceutical interest (73). The solubility in water of hydroxypropyl-βCD (HPβCD) and

randomly methylated βCD (RAMEB) is more than 600 and 500 mg/mL compared to 18 mg/mL

of the native βCD (82,83). The same behavior applies for HPγCD (>500 mg/mL) (82,83).

Optimum aqueous solubility is achieved when the number of methyl groups reaches 13-14; and

above that threshold the solubility decreases (83).

All the native cyclodextrins are classified as “Generally Regarded As Safe” (GRAS) by the FDA

(84). Hydroxypropyl β- and γ-Cyclodextrin (HPβCD and HPγCD) and sulfobutylether β-

cyclodextrin (SBE-β-CD) are listed in the FDA’s compiled list of inactive formulation ingredients

(85). Alfadex (α-CD), Betadex (β-CD), γ-cyclodextrin, SBE-β-CD, HPβCD, randomly methylated

β- cyclodextrin (RAMEB) are listed in the updated guidelines on the ‘Excipients in the labelling

and package leaflet of medicinal products for human use’ (SANTE-2017-11668) (86). The most

pharmaceutically appropriate β-CD derivatives are RAMEB with a 1.8 average number of

0.57 nm

1.37 nm 1.53 nm

0.78 nm

1.69 nm

0.95 nm

6 glucose units 7 glucose units 8 glucose units


15

substituents per repeated glucose unit and HPβCD with 0.65 (73,87). All the aforementioned

cyclodextrins have been approved and marketed by the European Pharmacopoeia, Japanese

Pharmaceutical Codex and the US Pharmacopoeia/National Formulary. Particularly, αCD can be found in intravenous alprostadil solution (Carveject® Dual from Pfizer). Native β- cyclodextrin

(βCD) can be found in numerous marketed oral, topical, buccal and rectal drug formulations:

notable examples are Nicorette® (sublingual tablets, Pfizer) and Nimedex® (tablets, Novartis).

Hydrocortisone solution (Dexocort, Actavis), indomethacin eye drop solution (Indocid, Chauvin)

and itraconazole oral and i.v. solutions (Sporanox®, Janssen) contain HPβCD and insulin nasal

sprays which contain RAMEB. Voltaren® Ophthalmic (Novartis) is an eye drop solution containing

diclofenac sodium salt and HPγCD (73).

Cyclodextrins have long been known for their ability to increase the water solubility of

hydrophobic drugs by entrapping them into their non-polar cavity (88). Long before that, the

purpose of using CDs in the pharmaceutical industry has been to protect drugs from oxidation

and heat, prolong their shelf-life and masking unpleasant flavours and odours (89).

Cyclodextrins have been investigated for the treatment of infectious diseases by formulating

them with anti-infective drugs. Their unique structure enables cyclodextrins to fully or partially

incorporate hydrophobic drugs in the cavity providing the drug with improved solubility properties

and enhanced biological activity. Table 1.3 shows a summary of antibiotic-cyclodextrin

complexes that have been studied and how the complex improved drug efficacy. In 2003, a US

patent was filed for improved growth inhibition of several Gram negative bacteria by HPβCD and RAMEB complexes with chlorhexidine, gentamycin, tobramycin and tetracycline via improved

permeation through the bacterial cell envelope (90). The inventors claimed that this patent was

of great interest for a wider range of antibiotics: penicillin derivatives, cephalosporins,

aminoglycosides, macrolides, rifamycin, chloramphenicol, imipenem, vancomycin, tetracyclines,

fusidic acid, novobiocin, neomycin, bacitracin, polymyxin, colistimethate, colistin and gramicidin.

Over the last few years, the interest in using drug-CD complexes to enhance antibiotic efficacy

is not based only on their enhanced permeation but also on improving drug water solubility,

stability and dissolution. For instance, native βCD and γCD protected penicillins from β-lactamase hydrolysis in E. coli and improved the cell permeation in vitro (91). Cyclodextrin

complexes have been investigated as drug release modifiers for sustained and prolonged drug

delivery, such as in the case of rifampicin, tobramycin and meropenem (92–95).


16

Table 1.3: Summary of antibiotic-cyclodextrin complexes investigated by research groups to achieve sustained drug release, better bioavailability, increased antimicrobial activity and enhanced drug solubility.

Cyclodextrin Drug Outcome Bacteria used Ref β-CD or γ-CD Penicillins Protection against β-

lactamase hydrolysis in vitro/ internalization into

cells

E. coli (91)

b-CD modified with

crosslinkers

rifampicin, novobiocin, vancomycin

Sustained drug release/ Increase bioavailability

S. aureus (92)

modified CDs rifampicin, tobramycin

Sustained drug release for hip-prosthesis

(93)

HPβCD fluoroquinolones Increased drug solubility by two to five times and

increased stability

(96)

βCD meropenem Enhance stability in aqueous solutions and in

the solid phase Sustained transfer/Constant drug concentration/greater

inhibition (bactericidal action)

P. aeruginosa, Rhodococcus equi, Listeria

ivanovii

(94)

HPβCD RAMEB

chlorhexidine, tetracycline, tobramycin, gentamicin

Improved biological activity (inhibit microbial growth/ eliminated toxic effects/

improved the dosing treatment)

P. gingivalis B. forsythus B. cereus

(90)

Triacetyl α-, β-, or γCD

vancomycin Prolonged drug release and maintained the antimicrobial activity of delivery systems for parenteral site-specific

administration

Staphylococcus aureus or

Enterococcus hirae

(95)

1.6.2 Liposomal carriers

The use of liposomal carriers has been regarded as a promising tool to enhance antibiotic activity, reduce drug toxicity and overcome bacterial resistance mechanisms (97,98). Liposomes are

universal carriers for both hydrophilic and hydrophobic compounds and offer a biocompatible,

biodegradable and non-immunogenic delivery system. Their unique physicochemical properties,

such as small and controllable size, large surface area and potential to target a specific site of

action via surface functionalization, enables them to improve drug pharmacokinetics and

biodistribution.

A wide variety of promising liposomal carriers for antimicrobial agents have been reported, which

are classified according to their lipid composition, membrane rigidity, surface properties and

ability to trigger drug release. These classes are: conventional, fusogenic and surface-modified

liposomes. Conventional liposomes are comparatively rigid carriers composed of phospholipids


17

with or without cholesterol, while fusogenic liposomes, also known as fluidosomes, are

characterized by more fluid lipid bilayers and their ability to fuse with other lipid membranes.

Table 1.4 presents several examples of how liposomes have been investigated and the outcome of the research in Gram negative antibacterial activity.

Liposomes can enhance the activity against both intracellular (Brucella melitensis, Salmonella

enterica serovar Typhimurium and Francisella tularensis) and extracellular (Pseudomonas

aeruginosa, Escherichia coli, Klebsiella pneumoniae and Burkholderia cepacia) pathogens (99).

Liposomal carriers designed for the eradication of intracellular pathogens are taken-up by cells of the mononuclear phagocyte system. For instance, conventional liposomes encapsulating

gentamicin lead to drug accumulation in the liver and spleen as a result of MPS cell phagocytosis,

resulting in decreased bacterial number in these tissues (99). Liposomes designed for targeting

extracellular pathogens can function in two main ways to aid drug delivery. Firstly, they can

entrap, dissolve or encapsulate the active ingredient or, secondly, they can attach, conjugate or

absorb the active substance. These properties can be applied to facilitate the administration of

antimicrobial drugs, thereby overcoming some of the limitations of antimicrobial drug uptake.

Table 1.4: Several liposomal carriers investigated against Gram negative bacteria.

Liposome composition Drug Bacteria Outcome Ref DPPC/Chol/SA,

DPPC/DPPG, DPPC/PI, DPPC/Chol/DDAB

Chlorhexidine P. vulgaris Enhanced bactericidal activity

Diffusion mechanism

(97)

DPPC/Chol (2:1) Amicacin P. aeruginosa Superior activity of inhaled liposomes than to free drug

(100)

DPPC/DMPG in several molar ratios (15:1, 18:1,

10:1)

Tobramycin P. aeruginosa B. cepacian

E. coli S. maltophilia

Superior activity of fluidosomes than free

drug Fusion mechanism

(101–103)

DPPC/DMPG (1:1) Gentamicin-gallium

P. aeruginosa Eradication of biofilm (104)

PC(60 mg/mL)/Chol Rifampicin High encapsulation and nebulization

efficiency Good stability

(105)

1.7 Aims and scope of this project

Most of the antimicrobial drugs in the current clinical pipeline are modified versions of existing

classes of antibiotics which are likely to offer only short-term solutions to the current AMR crisis

(106). A novel antimicrobial agent, PPA148 (belonging to the byrrolobenzodiazepine (PBD)

group of anticancer and antimicrobial drugs), has shown very promising activity against Gram

negative bacteria, and was synthesized at King’s College London (107). Among other

antimicrobial agents, PPA148 was included in a patent filed by Rahman et al. regarding its activity


18

against both Gram positive and negative bacteria (107). Its inhibitory concentration (MIC) in the

presence and absence of the efflux pump inhibitor PAβN (107,108) revealed promising activity

against Gram negatives amongst the ESKAPE pathogens. A possible drug efflux from the bacterial membrane was also observed because the MIC decreased in all bacterial strains tested

in the presence of the pump inhibitor (107). The focus of this project is on Gram negative bacteria

because of their increasing contribution to the incidence of ventilator-associated pneumonia

(VAP) (109), low OM permeability and high activity of multiple efflux pumps (29), which facilitate

resistance. Therefore, the aim of this project has been to develop a suitable carrier to overcome

the drug efflux and increase drug efficacy against Gram negative bacteria. Taking into account

the physicochemical properties of PPA148, the challenges of drug uptake across bacterial membrane and multi-drug resistance in Gram negative bacteria, a formulation based on a drug-

cyclodextrin complex incorporated into fluidosomes was selected.

1.7.1 Drug-in-Cyclodextrin-in-Liposomes

The system described as drug-in-cyclodextrin-in-liposomes allows delivery of hydrophobic,

insoluble and membrane-destabilizing drugs by combining the unique properties of liposomes,

with those of cyclodextrins (110). As mentioned above, each delivery system individually enhances drug bioavailability and stability (chemical and physical) and reduces toxicity; but their

combination offers a superior formulation (111).

This formulation was first introduced by McCormack and Gregoriadis to enhance the delivery of

dehydroepiandrosterone, retinol and retinoic acid (111). This approach has also been used in

the food industry to increase the stability and water solubility of natural molecules, decrease their volatilization, allowing controlled release, while maintaining the biological effects (antioxidant

activity) of free ones (112). Nerolidol, which is a natural sesquiterpene with antibacterial activity,

exhibited prolonged release and increased photostability when encapsulated as a complex with

HPβCD-in-liposomes, making it a strong candidate for a natural preservative use in the food

industry (113). Recent reviews reveal that a number of drugs have been encapsulated in

liposomes as inclusion complexes with cyclodextrins (CDs) (110,114). Particularly, this type of

formulation was tried with anti-inflammatory drugs (115,116) and anesthetics (117) to improve their encapsulation efficiency and membrane permeation; with anti-cancer drugs to improve their

solubility and prolonged retention (118,119); with Ca2+ channel blocker drugs to increase

entrapment and liposome stability (120); and with immunosuppressants to improve solubility,

drug uptake and penetration (121).

While there are reports of antibiotics solubilized by either inclusion complexes with CDs (Table 1.3), or encapsulated in liposomes (Table 1.4), we are not aware of any publication on antibiotics

formulated in CD complexes encapsulated into liposomes.


19

1.7.2 Plan of the study and techniques used

This work is an attempt to increase the water solubility and enhance the antimicrobial activity of

the novel antimicrobial agent, PPA148, against Gram negative bacteria by formulating it with cyclodextrins and liposomes. The first experimental chapter (Chapter 2) introduces the novel

antimicrobial compound, its synthetic pathway and presents the efforts used to enhance its water

solubility by using two types of modified βCD, HPβCD and RAMEB. The binding constant of a

1:1 complex is measured through the band shift method using fluorescence spectroscopy.

Fluorescence was studied as a function of host (CD) concentration at fixed guest (drug)

concentration. The increase in fluorescence intensity and blue shift in the spectra indicates the

formation of the host-guest inclusion complexes of drug with cyclodextrin. The overall

stoichiometry of the complex was better investigated measuring changes in the proton resonances using NMR.

The objective of Chapter 3 is to develop a robust method for the fabrication of the drug-in-

cyclodextrin-in-liposome formulation and compares its microbial activity with that of the free drug.

The encapsulation efficiency and loading capacity of the carrier is determined by spectroscopic

techniques. The formulated drug was used to challenge live bacteria to assess its antimicrobial efficacy and compare it with the efficacy of the unformulated drug.

Chapter 4 investigates the diffusion through lipid membranes of pure PPA148, PPA148/CD

complex and empty liposomes separately using interfacial techniques. Model lipid monolayers

are used to mimic the lipidic content of IM and OM of the Gram negative bacterial envelope, to

assess their interaction with drug and carrier components using the Langmuir trough technique. An asymmetric model Gram negative bacterial outer membrane is used to investigate the

interaction of the liposomal carrier with the lipidic content of the OM. Details on the interfacial

techniques used throughout this work is given in the next section (1.7.3).

Throughout this study, rifampicin and gentamicin sulfate were used as positive and negative

controls for the novel PP-A148 in the membrane interaction studies using monolayers at the air/liquid interface, whilst in bilayer studies chlorhexidine digluconate (CHD) was used as a

control drug (Table 1.5, Figure 1.9). Rifampicin and gentamicin are both broad spectrum

antibiotic agents against a series of Gram negative bacteria. The former passively diffuses

through the cell envelope, while gentamicin follows a self-promoted diffusion across the OM and

active transport across the IM (32,33,122). Chlorhexidine is known for its ability to damage

membranes (123).


20

Table 1.5: Physicochemical properties of PPA148, rifampicin, gentamicin and chlorhexidine.

Physicochemical properties

PPA148 Rifampicin Gentamicin Sulfate CHD

Molecular weight (g/mol)

698.25 822.41 477.6 897.76

LogP 0.76 2.7 -3.1 6.2 logS -7.461 - - -

S (mg/mL) - 2.5 100 50%w/v tPSA ( Å2) 142.11 217 200 455

Figure 1.9: Chemical structure of rifampicin, gentamicin and chlorhexidine.

1.7.3 Interfacial techniques used to investigate drug-membrane interactions

In the current work, the interaction of pure PPA148 and its formulated form with OM and IM Gram

negative bacterial membranes was investigated using two interfacial techniques: subphase

injection techniques using Langmuir monolayers and neutron reflectivity (NR) on substrate-

supported asymmetric bilayers. In both techniques, phospholipids and glycolipids (specifically lipopolysaccharides), were used to mimic the lipidic component of the IM and the OM,

respectively.

PE and PC possess zwitterionic headgroups and PG and CL are acidic lipids (Figure 1.3). The

relative proportions of these lipids and their generalized molecular shapes affect the way in which

they pack together in a membrane. PE has a conical shape (Figure 1.3) which imparts a large negative curvature to the membrane, CL also favors negative curvature, whilst PG and PC have

Rifampicin Gentamicin sulfate

Chlorhexidine


21

cylindrical shapes and are considered as low curvature lipids (124–126). E. coli lipid extract,

DPPC/DPPG and POPC/POPG mixtures were used to simulate the lipidic component of the

Gram negative IM. The composition of the lipid extract was 67% PE, 23.2% PG and 8.9% CL. White and co-workers (127) analyzed the same E. coli extract as the one used in this research

and determined the composition of fatty acids as well (Table 1.6).

Table 1.6: Fatty acid composition of E. coli lipid extract as it was found by White et al. (127).

R1 or R2: Fatty Acid Molecular Formula w/w% (127) C14:0 CH3(CH2)12CO- 1.5 C16:1 CH3(CH2)5CH=CH(CH2)7CO- 4.2 C16:0 CH3(CH2)14CO- 37.3 C17:0cyclo CH3(CH2)5CHCH2CH(CH2)7CO- 18.2 C18:1 CH3(CH2)5CH=CH(CH2)9CO- 31.1 C18:0 CH3(CH2)16CO- 0.9 C19:0cyclo CH3(CH2)5CHCH2CH(CH2)7CO- 6.7

Glycolipids and lipopolysaccharides (LPS) in particular, are a different category of lipids and were

used in this work to mimic the outer leaflet of the OM. In this study, we have used three different

types of LPS as shown in Figure 1.10 with different lengths of oligosaccharide chain. Rc J5 and

Ra EH100 LPS are extracted from the J5 and EH100 E. coli bacteria, respectively, while Re Lipid

A is extracted from R595 S. minnesota. Rc J5 LPS and Re Lipid A were used in the interaction studies on monolayers at the air/liquid interface of a Langmuir trough. The Ra EH100 LPS was

used to form the outer leaflet of a model asymmetric bilayer, which was used to examine the

interaction of fluidosomes with the model membrane by neutron reflectivity.

1.7.3.1 Langmuir trough isotherms

The Langmuir trough technique is used to form planar monolayers at the air-liquid interface and measure the partitioning properties and kinetics of molecules binding to biomimetic model

membranes. The most important indicator of the monolayer properties of an amphiphilic material

is given by measuring the surface pressure as a function of the area of water surface available

to each molecule. This is carried out at constant temperature and is known as a surface pressure

- area isotherm or simply Langmuir isotherm. Usually an isotherm is recorded by compressing

the film (reducing the area with the barriers) at a constant rate while continuously monitoring the

surface pressure (Figure 1.11). When the available area for the monolayer is large, the distance

between adjacent molecules is large and their interactions are weak. The monolayer can then be regarded as a two-dimensional gas (2D). Under these conditions, the monolayer has little

effect on the surface tension of water. If the available surface area of the monolayer is reduced

by a barrier system, the molecules start to exert a repulsive force on each other. This two-


22

Figure 1.10: Structure of Ra LPS derived from E. coli EH100, Rc LPS derived from E. coli J5 and Re LPS derived from S. minnesota.

dimensional analogue of a pressure is called surface pressure. At low surface pressures the

monolayers exist in the gaseous state (G) and upon compression might undergo a phase

transition to the liquid-expanded state (L1) (128). Upon further compression, the L1 phase

undergoes a transition to the liquid-condensed state (L2), and at even higher pressure the

monolayer finally reaches the solid state (S) (Figure 1.11). If the monolayer is further compressed

after reaching the S state, it will collapse into three-dimensional structures. The collapse is

generally seen as a rapid decrease in the surface pressure or as a horizontal break in the isotherm if the monolayer is in a liquid state.

The advantage of using such films is that they are of defined lipid composition and controlled in

terms of their surface area (129). P-Area isotherms show the relationship between the molecular

area of a film occupied on the liquid surface and the surface pressure observed. Langmuir films


23

were used as a convenient platform for studying the interaction of the novel compound with the

lipid monolayers at the air/liquid interface at 30 mN/m, which is widely accepted as a monolayer

surface pressure yielding lipid packing with molecular areas most accurately corresponding to those in biological membranes (130).

Figure 1.11: Langmuir trough lipid monolayers and their transition stages upon compression as described by changes in the Surface Pressure-Area isotherm. The illustration depicts the packing or the lipid alcyl chain upon compression from its 2D gas state until the collapse of the monolayer and the formation of 3D structures.

1.7.3.2 Neutron reflectivity

Neutron reflectivity is a more sophisticated way of investigating the composition and the relative location of molecules within and at the interface of supported asymmetric bilayers. Neutrons are

neutral particles which are characterized by their non-destructive nature, sensitivity to light

elements (hydrogen/deuterium, carbon, nitrogen and oxygen) and long penetration depths into

matter (131,132). Neutrons interact directly with the nuclei of molecules in a sample and are

scattered due to the presence of the strong nuclei observed in classical optics.

Neutrons behave as particle waves and thus follow in principle the quantum-mechanic rule

described by Planck and de Broglie (Appendix A). The kinetic energy of a neutron beam is related

to the wavelength of the periodic neutron wave by its momentum (particle mass and velocity,

EquationA1. 4). Modifications of neutron momentum p and their energy (E) indicate changes in

the neutron track (Figure 1.12).


24

Figure 1.12: The classic optics of a beam, hitting a planar surface, related to the geometry of this process with the wavevector transfer. The wavevector is the “spatial” frequency of the neutron wave and is described by the |𝑘| = @A

B , where k is the wavevector and λ is the neutron

wavelength. 𝑘CDDD⃗ , 𝑘FDDDD⃗ and 𝑘GDDDD⃗ are the wavevectors of incident, reflected and refracted neutron beam, respectively. 𝑄D⃗ is the wavevector transfer (𝑘CDDD⃗ − 𝑘FDDDD⃗ ), 𝜃HI, 𝜃JKL and 𝜃G are the incident, outgoing and refracted angle of the neutron beam.

In an ideal simple scattering experiment, the scattering is elastic and, therefore, there is no

energy transfer to be considered. The scattering event is restricted to the wavevector transfer

(𝑄 = |𝑘H| − |𝑘F|) (Figure 1.12). The condition of “no energy exchange” (𝐸H = 𝐸F) between the

neutron and the sample implies that there is no modification of the wavelength of the neutron

(|𝑘H| = |𝑘F|). Based on elementary trigonometry (Figure 1.12) leads to the following wavevector

transfer formula:

𝑄 = OA PQRSTUB

Equation 1.1

, where Q describes the wavevector transfer position in reciprocal space in the z direction (Figure

1.13B), 𝜃HIis the incident angle and λ the wavelength of the neutron bean. Reflected neutrons

from a thin film may undergo strong interference depending on the wavelength of the neutron, their state of polarization, the thickness of the layer and the neutron refractive indices of the

media involved (Figure 1.13). The scattering length densities (SLD) reflects the magnitude of

scattering by each molecules in the sample, and is similar to a refractive index in light scattering.

The relative values of SLDs of the constituting molecules is of major importance for the analysis

of complex structures.

A standard multilayer method is used to calculate the reflectivity profile. Neutrons travel through the matter at different speeds and, thus, penetrate the layers to different depths (Figure 1.13A).

The difference between the wavevector transfer of the two ejected neutrons (k), which appears

as interference fringes in the reflectivity profile (Figure 1.13A), is directly related to the thickness

of the interfacial layer by:

𝑘 = 𝑄@ − 𝑄V =@AW

Equation 1.2

, where 𝑄@ − 𝑄V describes the separation of between the interference fringe and d is the thickness

of the interfacial layer. Although the wavevector transfer (Q) links the magnitude of momentum


25

transfer with the scattering angle, it does not capture all the spatial aspects. The scattering of a

neutron occurs in a three-dimensional system; therefore the position of the scattered wavevector

𝑘F is defined by a second angle φ (Figure 1.13B), which is required for the measurement of

rotation about 𝑘H. The angles θ and φ determine the direction of the outgoing neutron, the same

way as longitude and latitude give the position of a point on the surface of the Earth.

Figure 1.13: (A) The oscillations in the reflectivity profile resulting from the interference which depends on the thickness (d) of the layer (in terms of the position) and the difference in scattering contrasts between the respective interfaces (in terms of the amplitude). 𝑘LDDD⃗ is the wavevectors of the incident neutron beam and 𝑘FVDDDDDD⃗ and 𝑘F@DDDDDD⃗ the outgoing neutron beam from the top and bottom of the interfacial layer, respectively. 𝑛V and 𝑛@ are the refractive index of light while SLD1 and SLD2 are the refractive indexes of neutrons in two different bulk phases. 𝜃HI, 𝜃J and 𝜃G are the incident, outgoing and refracted angle of the neutron beam. (B) The scattering geometry in both Cartesian and spherical polar coordinates with ϕ and θ are angles determining the direction of the outgoing neutron beam. 𝑘FDDDD⃗ is the wavevectors of reflected neutron beam and λ is the neutron wavelength.

The SLD profile is directly related to the distance along the z axis and gives information about

the thickness, roughness and solvation of a particular layer. The use of contrast variation is

essential in NR experiments in order to obtain as much information as possible about each layer. A way to achieve this is the use of isotopic substitutions in the whole molecule or part of a

molecule, to manipulate contrast (neutron refractive index). The SLD is related to the refractive

index by the following formula (133):

𝑛 ≈ 1 − YBZ

@A Equation 1.3

, where n is the refractive index, ρ is the SLD and λ is the neutron/light wavelength. The

possibility to selectively labelling a molecule with isotopes of different SLD allows the neutron beam to highlight or mask specific regions of the molecule. One of the most extensively used

contrast variations is between hydrogen and deuterium (134,135), since the scattering length

densities of these two isotopes are markedly different.

Examples of contrast variation are described in Table 1.7. If tail-deuterated lipids are spread on

the air/D2O interphase in pure D2O, the total scattering intensity is given by the lipid headgroup

Incident neutron beam Reflected neutron beam

θr d

n1 (SLD1)

n2 (SLD2)

!"!#$ !#%

&" &'

k (

)

*

!"

!# =2-.

&/

(A) (B)


26

(135). It is possible to match the SLD of an H2O and D2O mixture to the scattering length density

of specific biological molecules and materials. For instance, a mixture of approximately 38% D2O

in H2O (v/v) matches the SLD of silicon, whereas a mixture of 13% D2O in H2O (v/v) would match the SLD of hydrogenated lipids. The matched molecules would not contribute to the neutron

scattering and the scattering profile contains no information regarding these molecules. In

air/water Langmuir interface experiments, it is convenient to use 8% D2O in H2O (v/v) solution

which has no scattering (SLD=0) (136). The SLD is identical to that of air and, therefore, the

solution is called null reflected water (NRW). In contrast, in supported lipid bilayers on a silicon

surface, a mixture of 32% D2O and 62% water matches the SLD of silicon and is called silicon

matched water (SMW).

Table 1.7: Potential isotopic contrasts and main information content of a monolayer on an air/water interface or a deposited lipid bilayer on a silicon surface consisting for tail deuterated phospholipids. NRW and SMW stand for null reflective water (8.2% D2O) and silicon match water (38% D2O) respectively and are used to cancel out the background of the two mediums and highlight the interface.

Contrast cartoon/ Mololayer

Main information

Contrast cartoon/ Bilayer

Main information

Lipid hydrocarbon

tail

Hydrocarbon lipid tails

Lipid headgroup

Lipid Headgroups

Lipid components

Interface Lipid bilayer

SLDD2O = 6.35×10−6 Å −2 SLDNRW = 0 Å −2

SLDH2O = -0.56×10−6 Å −2 SLDSMW = 2.00×10−6 Å −2

All these advantages have made neutrons an ideal tool for studying the composition and structure of supported model lipid membranes by measuring the diffracted neutron beam. In this

study, specular neutron reflectivity is used to characterize the in- plane model asymmetric

membranes and study changes in their structure caused by exposure to a liposomal drug carrier.

Ra-EH100 LPS and DPPC, which account for the lipidic component of the outer and inner leaflet

respectively of the OM, were deposited on a flat silicon surface forming a model asymmetric

membrane as presented in Figure 1.14.


27

Figure 1.14: Asymmetric lipid bilayer formed of Ra-EH100 LPS and DPPC and deposited on a silicon surface. The red arrows show the neutron path when the bean hits the surface.

A highly collimated beam of neutrons hits the flat surface and the neutrons travel through the

individual layers of the membrane. The intensity of the reflected radiation is measured as a

function of angle or neutron wavelength and plotted against neutron momentum. This reflectivity

data sets (all three contrasts) are then transformed into SLD vs distance and these data are fitted

to a mathematical layered model, providing a real space distribution of the data sets.

Neutron Beam

Neutron scattering

from each layerInner Headgroup

Outer Headgroup

Outer h-TailsInner d-Tails

Silicon Block

Solution

Silicon Oxide

Synthesis, characterization and solubility of PPA148

28

Chapter 2 Synthesis,

characterization and solubility of

PPA148


29

2.1 Introduction

Pyrrolobenzodiazepine (PBD) drugs are a family of natural anticancer-antibiotics isolated from

various Actinomycetes bacteria. They are a class of sequence-selective DNA minor-groove

agents. The first PBD monomer, anthramycin, was originally discovered as a metabolite of

Streptomyces refuineus (137). Since then, several natural PBDs have been isolated from

Streptomyces, Micrococcus (138) and Klebsiella (139) bacteria and various PBDs have been chemically synthesized (140). A tricyclic scaffold is the basic skeletal structure of the PBD

molecules (Figure 2.1) (107,108,141,142). The system consists of a substituted aromatic A-ring,

a diazepine B-ring and a pyrrolidine C-ring (Figure 2.1). Their biological activity is facilitated via

covalent binding of the imine moiety at position N10-C11 to the C2-amino group of guanine. The

imine moiety at N10-C11 is the electrophilic center responsible for DNA binding (142). It has

been reported that PBDs have antimicrobial activity against Gram positive and Gram negative

bacteria. However, their high degree of toxicity has hindered their use as antibiotics (108). Derivatization and the use of different synthetic routes from several research groups have

attempted to decrease their toxicity while maintaining their antibacterial activity (108).

Figure 2.1: Mechanism of tricycle PBD core binding to the N2 of guanine in the DNA minor groove reproduced from Brucoli et al. (143).

A novel series of modified PBD biaryl hybrids with lower cytotoxicity against eukaryotic cells compared to the monomer was chemically synthesized at King’s College London (108). The

chemical structure of this class is described as a tricycle PBD scaffold linked to a lateral

polyamidic tail in position 8 via a 4-carbon aliphatic chain. PPA148 is the lead antimicrobial

compound of this novel class of molecules. Its tail comprises three units linked with an amide

bond, an N-methyl-pyrrole-4-amino methyl ester, a benzofuran unit and a thiomorpholine unit

(Figure 2.2). Three building blocks have been proven to give optimum antibacterial activity (108).

A screening of the physicochemical properties was performed using ChemDraw software to obtain a first indication of the lipophilicity and polarity of PPA148. Based on the structure, drug

family and physicochemical properties estimated by ChemDraw (Mw = 698.25 g/mol, clogP =

2.09 and logP = 0.76) (Table 2.1), PPA148 is a large and non-ionic lipophilic molecule, whose

bactericidal activity is attributed to its binding onto bacterial DNA. According to O’Shea et al. (12)


30

and Brown et al. (144), antimicrobial agents have distinct properties, which often violate Lipinski’s

rule of five (Table 1.1) (12,13) and depend on their activity against Gram positive and/or Gram

negative bacteria due to differences in the constitution of the bacterial cell envelope as presented in Table 2.1. Based on these findings, PPA148 seems to have a similar lipophilicity (clogP) to

antimicrobials acting against Gram positive, and similar polarity (tPSA) to those acting against

Gram negative bacteria (Table 2.1). Nevertheless, measurements of the minimum inhibitory

concentrations (MIC) showed that PPA148 is active against both types of bacteria (Appendix B,

Table B 1; Table B 2) but present reduced permeability in Gram negative. In fact, MIC

experiments were conducted in the presence of the efflux pump inhibitor, PAβN, which showed

significantly reduced MIC values, thus indicating that PPA148 is affected by the efflux pump. The low water solubility and the results from the microbiological assay suggest that enhancing the

solubility of the drug may improve its uptake and thus help increase its antimicrobial activity

against Gram negative bacteria.

Figure 2.2: Chemical structure of PPA148.

Table 2.1: Comparison of the physicochemical properties of PPA148 estimated by ChemDraw compared (the experimental values are presented in parenthesis) with the generic optimal properties of antibacterial drugs for oral or parental administration as calculated by O’Shea et al. (12) based on data collected from the literature and drug data bases.

Physicochemical properties PPA148 Optimal properties

against Gram negative Optimal properties

against Gram positive

Mw (g/mol) 698.25 414 (cutoff of 600) 813 LogP 0.76 - - cLogP 2.09 -0.1 2.1 LogS -7.461 (-4.37) - tPSA ( Å2) 142.11 165 243

Mw: Molecular weight

The approach to enhance water solubility is to use cyclodextrins, which are well-known

pharmaceutical excipients, capable of forming inclusion complexes with a wide range of poorly soluble molecules (72,144) (Figure 2.3). Cyclodextrins are cyclic oligosaccharides containing 6

(αCD), 7 (βCD) or 8 (γCD) (α-1,4)-a-D-glucopyranose units, respectively. All hydroxyl groups are

oriented to the cone exterior with the primary hydroxyl groups at the narrow side of the toroid


31

cone and the secondary on the opposite (wider) side (145) (Figure 2.3). The hydrogens of the

skeletal carbons at position 1, 2 and 4 are mainly located on the exterior side of the toroid

molecule and the hydroxyl groups at position 2, 3 and 6 are oriented toward the exterior of the cone. The central cavity includes the hydrogens at position 3 and 5 and the ring of glycosidic

ether oxygen is present with H-6 located near the cavity (145) (Figure 2.3). This central cavity

has a polarity similar to ethanol (146). Due to this unusual structure, cyclodextrins are able to

encapsulate (partially or wholly) a wide variety of guest molecules of suitable size, shape,

structure and physicochemical properties, resulting in a stable association without the formation

of covalent bonds (77,87). In an aqueous environment, the hydroxyl groups form hydrogen bond

with water molecules, creating a hydration shell around the CD molecule.

This capability of CDs to form water-soluble complexes with (lipophilic) poorly soluble drugs has

been used extensively to increase their solubility (73,147). Two types of complexes can be

formed: inclusion and non-inclusion. With inclusion complexes, the central cavity provides a

lipophilic microenvironment to the molecule; the guest (drug) and host (CDs) are in dynamic

equilibrium with the complex (Equation 2.1). Non-inclusion complexes are formed by hydrogen bonds between the outer surface of the CDs (hydroxyl or another substituted group) and the drug

(73). It is possible that both types of complexes can co-exist (148,149).

Figure 2.3: Different representations of the structure and conformation of β-cyclodextrin, showing the orientation of the hydrogens and hydroxyl groups. Native βCD contains hydroxyl groups, while its derivatives can be substituted by different functional groups. RAMEB contains either hydrogens or methyl groups, while HPβCD has either hydrogens or hydroxypropyl groups.

The complexation process is characterized by the equilibrium or binding constant (K) (Equation

2.2). The binding constant provides a measure of the affinity between the guest and the host.

The following equations describe the interaction and the stoichiometry of the complexes formed.

𝑚𝐶𝐷 + 𝑛𝐷 ⟷𝐶𝐷_𝐷I Equation 2.1

𝐾 = abcbUabc×bU

Equation 2.2


32

, where m and n describe the stoichiometry of the CD/drug complex. For instance, a 2:1 CD/drug

complex would have m of 2 and n of 1. Binding constants for cyclodextrin lie between 50 and

2000 M-1 (73) and inconsistencies among the methods used to determine the binding constant have been reported (75).

In this chapter we report the synthesis of a novel antibiotic, PPA148, its characterization and the

formation of inclusion complexes with either HPβCD or RAMEB to increase its solubility. A

convergent synthetic route was applied based on the separate synthesis of the tricyclic PBD

scaffold and the lateral polyamidic tail (107,108). The two final pure core and tail molecules were linked via an amidic bond to form the final PPA148. Following the synthesis, the physicochemical

characteristics, such as solubility, of this novel antimicrobial agent were measured. Finally, the

drug was formulated with HPβCD and RAMEB and drug/CD binding constants measured using

fluorescence spectroscopy. Given the low solubility of the drug, this was the first step towards

potentially enhancing its efficacy by forming inclusion complex with a βCD derivative.

Enhancement of water solubility will help to optimize the drug for local use to the lungs by

avoiding precipitation and side effects (150). The second step for achieving increased antibacterial activity was to prepare a liposomal carrier to help PPA148 breach the bacterial cell

envelope and bind to the chromosomal DNA (Chapter 3).

2.2 Materials

Reagents and chemicals used for the synthesis: (Diacetoxyiodo)benzene (BAIB), 3,4-

dihydro-2H-pyran (DHP), 4-(dimethylamino)pyridine (DMAP), ptoluenesulfonic acid

monohydrate (PTSA) (≤ 100%, sulphuric acid 1 - 5%), 2,2,6,6-tetramethylpiperidinooxy

(TEMPO), tetrakis(triphenylphosphine)palladium(0), triphenylphosphine and N-(3-

dimethylaminopropyl)-N’-ethylcarbodiimide hydrochloride (EDCI) were purchased from Sigma-

Aldrich, UK. Dichloromethane (DCM), dioxane, N,N-dimethylformamide (DMF), methanol,

acetone, ethyl acetate (EA), acetonitrile (ACN) and thiomorpholine were purchased from Sigma-

Aldrich, UK. LC/MS grade water was supplied by Merck, UK. Ethyl-5-amino-1-benzofuran-2-

carboxylate and 4-tert-butoxycarbonylamino-1-methyl-1H-pyrrole-2-carboxylic acid were purchased by Fluorochem Ltd (Hadfield, UK). Citric acid, sodium chloride (NaCl), brine and

sodium bicarbonate (NaHCO3) were purchased from Sigma-Aldrich, UK. Formic acid for use in

LC/MS was purchased by Fisher Scientific UK. Silica was purchased from Sigma-Aldrich, UK,

and TLC silica gel 60 F254 aluminium sheets (20 cm) were supplied by VWR International, UK.

Deuterated water, chloroform and DMSO were purchased by Sigma-Aldrich, UK.

Cyclodextrins: Hydroxypropyl-β-cyclodextrin (HPβCD), randomly methylated β-cyclodextrin

(RAMEB) and heptakis (2,6-di-O-methyl)-β-cyclodextrin (DIMEB) were purchased from Sigma-

Aldrich, UK


33

Active Pharmaceutical Compounds: Rifampicin, gentamicin sulfate and chlorhexidine

digluconate were purchased by Sigma-Aldrich, UK.

Buffer preparation: Trizma(R) hydrochloride (BioUltra, for molecular biology, >=99.0%) and

HEPES (>=99.5%) purchased from Sigma-Aldrich, UK. Sodium chloride (NaCl), calcium chloride

(CaCl2) and magnesium chloride (MgCl2), supplied from Sigma-Aldrich, UK. The ultrapure water

at 18.2 MΩ was produced by a Purelab Ultra machine from ELGA process water (Marlow, UK).

2.3 Methods

This section describes the methods applied for the synthesis and characterization of all intermediate and final compounds. All the compounds were tested with TLC (2.3.2), LC/MS

(2.3.3), proton and carbon NMR (2.3.4) and FTIR (2.3.5). The conditions of these tests are stated

in separate sections from that of the synthetic pathway (2.3.1). Methods used for further

characterization of the final compound (aqueous solubility) and complex formation for enhancing

water solubility will be described at the end of this section (2.3.6: Adsorption assay using the

Langmuir trough, 2.3.7: Ultra Violet Spectroscopy, 2.3.8: Turbidimetric Assay, 2.3.9: Water

Solubility Test, 2.3.10: Quantification of drug binding affinities to Cyclodextrin (Fluorescence

Binding constant assay), 2.3.11: Phase solubility diagram, 2.3.12: Continuous method variation (Job’s plot))

2.3.1 Synthetic pathway of PPA148

The complete reaction sequence is shown in Scheme B 1, Appendix B (107). In the present work,

the synthetic route was performed according to Scheme 2.1. It is separated into 3 sections. The

first (Scheme 2.1A) describes the core synthesis, the second (Scheme 2.1B) the tail and the last (Scheme 2.1C) shows the coupling of the two parts to form the final compound #12 (PPA148).


34

Scheme 2.1: Synthesis of C8-linked pyrrolobenzodiazepine-carboxylate conjugate (A: PBD corecarboxylate, B: Tail conjugate and C: final core and tail coupling and deprotection steps)

Synthesis of core compound #7 (IUPAC name: allyl 11-hydroxy-7-methoxy-8-(4-methoxy-4-

oxobutoxy)-5-oxo-2,3,11,11a-hexahydro-1H-pyrrolo[2,1-c][1,4]benzodiazepine-10(5H)-carboxylate) was achieved by the following method. Compound #6 (1.73g, equi. 1) underwent a

mild aerobic oxidation by adding TEMPO (65.85 mg, equiv. 0.11) and diacetoxyiodobenzene

(BAIB, 1.48 g, equiv. 1.2) in dichloromethane (88 mL). The reaction was left under stirring for 12

hours with regular TLC monitoring (EA as mobile phase). The reaction mixture was washed with

saturated sodium metabisulphite (44 mL) followed by saturated aqueous NaHCO3 (2 × 44 mL),

water (44 mL) and finally brine (44 mL). The resulting solution containing crude #7 was dried

over MgSO4 and the solvent was removed by rotary evaporation under reduced pressure. The

crude product was purified using flash chromatography (EA/DCM).

N

BHcoN

O

OH

+H2N

O O

OEt

EDCI, DMAPDMF

N

BHocN

O

NH

O O

OEt

N

BHocN

O

NH

O O

N

S

1. NaOH, MeOH2. EDCI, DMAP, DMF, Thiomorpholine

#11

#10

MeOH4M HCl in Dioxane

N

H2N

O

NH

O O

N

S

De-protected #11

(A)

(B)

(C)


35

The synthesis of core compound #8 (IUPAC name: allyl 7-methoxy-8-(4-methoxy-4-oxobutoxy)-

5- oxo-11- (tetrahydro-2H -pyran -2 -yloxy)- 2,3,11,11a hexahydro 1H-pyrrolo [2,1c] [1,4]

benzodiazepine-10(5H) -carboxylate) was carried out by the following method. The N10-alloc protected carbinolamine #7 (1.3 g, equiv. 1) was added to a solution of DHP (2.6 mL, 10

equivalents) and a catalytic amount of PTSA (11 mg, equiv. 0.06) in ethyl acetate (24 mL). After

two hours of stirring, TLC analysis (DCM/EA (90:10) as mobile phase) showed completion of

reaction and the reaction mixture was diluted with ethyl acetate (23 mL). The resulting solution

was washed with saturated aqueous NaHCO3 (20 mL), followed by brine (20 mL). The ethyl

acetate layer was then dried (MgSO4) and evaporated using a rotary evaporator (RC 600 with a

SC920 G vacuum pump system and a C 900 chiller, KNF Lab (Oxford, UK), under reduced pressure. Product #8 was purified by flash chromatography (DCM:EA 90:10).

Core compound #9 (IUPAC name: 4-(10-(allyloxycarbonyl)-7-methoxy-5-oxo-11(tetrahydro-2H-

pyran-2-yloxy)2,3,5,10,11,11a-hexahydro1H pyrrolo[2,1c][1,4]benzodiazepine-8-yloxy)butanoic

acid) was synthesized using the following steps. Excess NaOH 0.5M was added to a solution of

#8 in dioxane at room temperature. The reaction mixture was left under stirring for 4 hours at which point TLC showed completion of the reaction. Dioxane was evaporated under vacuum and

the residue was diluted with water. The resulting solution was acidified with 1 M citric acid

followed by extraction with ethyl acetate (2 × 100 mL). The combined organic layers were washed

with brine (100 mL), dried over MgSO4 and finally concentrated using a rotary evaporator under

reduced pressure. Product #9 was purified by flash chromatography (DCM:EA 90:10).

The synthesis of the tail compound is a series of three amide coupling reactions. The tail

compound #10 (IUPAC name: 4-((tert-butoxycarbonyl)amino)-1-methyl-1H-pyrrole-2-

carboxamido)benzofuran-2-carboxylic acid) was synthesized according to the following

procedure. 4-((tert-butoxycarbonyl)amino)-1-methyl-1H-pyrrole-2-carboxylic acid (900 mg,

equiv. 1.2) was dissolved in DMF (4-5 mL). EDCI (598 mg, equiv. 2.5) and DMAP (558 mg,

equiv. 3) were added to the solution and was left under magnetic stirring in a pure N2 atmosphere

for 20 minutes. At this point, ethyl 5-amino-1-benzofuran-2-carboxylate (300 mg, equiv. 1) was

added to the reaction and left under magnetic stirring pure N2 atmosphere overnight. EA was added (30 mL) and then the organic phase (DMF and EA) was extracted with citric acid 0.1 M

aqueous solution (15 mL), saturated NaHCO3 (15 mL) and brine (15 ml). The collected organic

phase was dried with MgSO4 and subsequently evaporated under reduced pressure using a

rotary evaporator giving the crude compound #10. The crude was purified by column

chromatography using silica as a stationary phase and DCM/EA as a mobile phase to give the

final pure compound #10.

The next intermediate tail compound, #11, (IUPAC name: 1-methyl-5-((2-(thiomorpholine-4-

carbonyl)benzofuran-5-yl)carbamoyl)-1H-pyrrol-3-yl)carbamate) was made following the steps

described below. Compound #10 (369 mg, equiv. 1) was dissolved in MeOH/dioxane. Excess

NaOH 1M (equiv. at least 5) was added to the mixture and left under magnetic stirring at room


36

temperature overnight. Disappearance of the starting material, shown by TLC, led to completion

of hydrolysis. MeOH/dioxane was evaporated under reduced pressure using a rotary evaporator.

Citric acid 1M was added until pH became acidic (~2) causing formation of precipitate (acid). The precipitate was extracted with EA, the organic phase was collected and the solvent was

evaporated using rotary evaporator. The acid produced (equiv. 1.2) was dissolved in DMF (4-5

mL). EDCI (equiv. 2.5) and DMAP (equiv. 3) were added to the solution which was left under

magnetic stirring (in N2 atmosphere) for 20 minutes. At that point, thiomorpholine (equiv. 1) was

added to the reaction and left under magnetic stirring overnight. EA was added (30 mL) and then

the organic phase (DMF and EA) was extracted with citric acid 0.1 M aqueous solution (15 mL),

saturated NaHCO3 (15 mL) and brine (15 ml). The collected organic phase was dried with MgSO4 and subsequently evaporated using a rotary evaporator giving the crude compound #11. The

crude compound was purified by column chromatography using DCM:EA (90:10) as a mobile

phase to give the final pure compound #11.

The synthesis of the final compound #12, also called PPA148, included the first amide coupling

of compound #9 and #11 followed by the deprotection of the diazepine B-ring of the PBD scaffold with the formation of the imine bond. Pure compound #11 (100 mg) was de-protected by being

dissolved in MeOH (6 mL) and HCl 4M in dioxane (6 mL) at a ratio of 50:50 and left under stirring

for 2 h. The reaction was monitored with TLC. The reaction mixture was evaporated using a

rotary evaporator and the intermediate de-protected #11 was formed. The protected PBD

derivative (compound #9, equiv. 1.2) was dissolved in DMF and EDCI (equiv. 2.5) and DMAP

(equiv. 3) were added to the solution. The mixture was left under magnetic stirring (in N2

atmosphere) for 20 minutes. At that point, the de-protected compound #11 (equiv. 1) was added

to the reaction and left under magnetic stirring overnight (in N2 atmosphere). EA was added (30 mL) and then the organic phase (DMF and EA) was extracted with citric acid 0.1 M aqueous

solution (15 mL), saturated NaHCO3 (15 mL) and brine (15 ml). The collected organic phase was

dried with MgSO4 and subsequently evaporated using a rotary evaporator giving the intermediate

crude compound #12. The crude was purified by column chromatography using DCM:Acetone

(90:10 to 60:40) as a mobile phase.

The final compound was formed by de-protecting the pure intermediate compound #12. This

intermediate (equiv. 1) was dissolved in DCM. Tetrakis Pb (equiv. 0.05), triphenylphosphine

(equiv. 0.25) and pyrrolidine (equiv. 1.2) were added and the reaction kept under stirring for 20

min. The completion of the reaction was monitored with TLC. At that point the solvent was

evaporated using a rotary evaporator and high vacuum giving the crude of the final compound

#12. The crude was purified by column chromatography using DCM:Acetone (90:10 to 60:40) as

mobile phase.


37

2.3.2 Thin Layer Chromatography (TLC)

TLC is a planar chromatographic technique extensively used in synthetic organic chemistry as a

rapid and straightforward tool to determine each compound’s purity and a means of following the progress of a reaction. It was performed for all intermediates and the final compound. As a mobile

phase a mixture of organic solvents was used depending on the nature and affinity of each

compound (DCM:EA at different ratios for the intermediate and DCM:Acetone for the final

compound). Silica gel-coated aluminium sheets were used as a stationary phase. The mobile

phase travels through the stationary phase and carries the components of the mixture with it.

The movement of the analyte can be seen under the UV light and expressed by the retardation

factor (Rf). Rf depends on the distribution coefficient (K) and is the relation between the distance

travelled by the analyte from its origin (Kα) to the distance travelled by the solvent from its origin (Ks) using the following formula:

𝑅F =fgfh Equation 2.3

If the analyte spot is not UV visible, a number of solutions can be applied on the dried TLC plate

and produce colored products. Potassium permanganate (KMnO4) was used for all compounds

that can be oxidized such as phospholipids and cyclodextrins. UV radiation at 254 nm was used

for all samples.

2.3.3 Liquid Chromatography-Mass Spectrometry (LC/MS)

LC-MS was conducted with the triple quadrupole Agilent Infinity Automated LC/MS Purification

System to identify the purity and mass of the compounds at every step of the synthetic route.

The physical separation was performed on a OnyxTM monolithic C18 column with dimensions

of 50mmx4.60mm and pore size of 130 Å. The mobile phase was a mixture of HPLC grade water

with 0.1% formic acid (Solvent A) and HPLC grade ACN with 0.1% formic acid (Solvent B). The

gradient was ramped from 95% A at t = 0min to 50% A at 2 min, then to 10% at 2.5 min, and kept

it until 4.5 min and finally ramped to 95% at 4.6 min and keep there for 5 min. The injection volume was 10 µL, the flow rate was 1 mL/min and minimum/maximum flow ramp up/down was

100 mL/min2. The LC detector was an Agilent 1260 Infinity II Diode Array Detector (DAD) G7115A

and the detection wavelength range was 220-400 nm. The UV absorbance was recorded at 260

nm with a slit width of 4 nm at 0.8 °C. The LC was coupled with the mass spectrometer via an

electrospray ionization (ESI) interface operating in both positive (ESI+) and negative (ESI-)

modes. The nebulization pressure was 50 psig, the gas used N2 and the quadrupole temperature

set at 0 °C. The voltage capillary was 4000 V and 3500 V for ESI+ and ESI- respectively and the drying gas temperature and flow were 350 °C and 13 L/h. The sample cone (skimmer) voltage

and ion energy was set at 8 and 5 V respectively for both ionization modes. Lens1, Lens2, IRIS

lens and HED were set at 2.4, 23, -400 and 10000 V for the ESI+ and -4, 23, 400 and 10000 V

for the ESI-.


38

For all the compounds, the relative purity was assessed based on the LC chromatogram because

there were no analytical standards to assess their absolute purity. Therefore, it was calculated

based on the relative area of each intermediate and final compound as a percentage of all other peak areas in the chromatogram. For the purpose of this research, the term relative purity will be

referred to as purity or RP.

2.3.4 Nuclear Magnetic Resonance (NMR) spectroscopy

Liquid state NMR spectroscopy was carried out on an AscendTM 400 MHz spectrometer

equipped with a SampleXpress autosampler system (Bruker, UK) for all intermediates and the final compound of the synthetic procedure to identify their molecular content and purity. 1-

dimentional (1D) proton (1H) and carbon (13C) NMR were performed to assign the resonances.

The spectra were recorded using TOPSpin software for data acquisition and MestReNova

software for data analysis. 1H NMR parameters included sweep width of 8012.82 Hz, acquisition

time of 4.09 sec and 16 scans and were collected with a zg30 pulse program. 13C NMR used a

sweep width of 24038.46 Hz, acquisition time of 1.36 sec and 1024 scans. The solvent used was

either deuterated chloroform (CDCl3) or deuterated DMSO, depending on the solubility of the

compound.

2.3.5 Fourier Transform Infra-Red (FTIR) Spectroscopy

FTIR was carried out to confirm the presence of the basic functional groups of newly synthesized

compounds. It is an additional technique to NMR to confirm the structure of each intermediate

and the final compound. A Perkin Elmer, UK, Frontier FTIR (serial number 95462) with a LiTa03

detector, IR-laser wavenumber of 15798 cm-1 and a zinc selenide (ZnSe) crystal was used to analyze the samples in their solid state. For a better contact of the sample with the crystal, the

DATR 1 bounce diamond accessory with a pressure arm force indicator (part number L1250240,

Perkin-Elmer Ltd, UK) was attached to the main piece of equipment. All spectra were produced

using 16 scans and collected at a scan speed of 0.2 cm-1/min and spectral resolution of 2 cm-1.

The data were analyzed using Spectrum One software (version 10.03.06, Perkin-Elmer Ltd, UK).

Each spectrum was scale-base normalized by dividing the absorbance of each point with that of

the C-H stretch at around 2950 cm-1 (internal standard) to minimize the errors caused by the

amount of sample used.

2.3.6 Adsorption assay

The adsorption of the drugs chlorhexidine digluconate, rifampicin, gentamicin and PPA148 at the

air/water interface was recorded by measuring changes in surface pressure after subphase

injection. A Langmuir trough (NIMA Technologies Ltd., Coventry, UK) equipped with a calibrated

NIMA PS4 pressure sensor and a Wilhelmy plate (Whatman, grade 1, chromatographic paper), connected to a controlling computer with the NIMA IU4 interface unit. For all the experiments a

50 mm diameter perfluoroalkoxy (PFA) petri dish (Saint-Gobain Performance Plastics) with a 20


39

mL volume capacity, was placed over a magnetic stirring plate and referred to as the “small

trough”. A dust and contaminant free subphase (approximately 20 mL filtered ultrapure water

containing 0.9% w/v NaCl) was poured into the petri dish and a clean Wilhelmy plate was suspended from the NIMA PS4 pressure sensor and partially submerged into the subphase. The

subphase surface was cleaned by sweeping and suction using a pressure pump (Aspirator A-

3S, EYELA, Tokyo Rikakikai Co, Ltd) in order to remove any remaining dust and contaminants

from the surface (Π = 0± 2 mN/m).

Drug solution was injected (0.1 μL) below the surface directly into the subphase using a 1 mL disposable plastic syringe (fitted with a 25 G x 25 mm needle) of 1 mL volume (BD biosciences

UK, Oxford, UK) whilst continuously stirring. Chlorhexidine digluconate (CHD) and gentamicin

sulfate were dissolved in water whilst rifampicin and PPA148 were dissolved in DMSO. In each

case, the final drug concentration in the trough was 2 μg/mL. The adsorption of the drug on the

surface causes changes in surface pressure which are recorded over time whilst maintaining a

constant surface area (21.3 A2/m) at 23 °C. Each sample was run in triplicate (n=3) and the

changes in surface pressure were plotted against time (Π and A-Time isotherm). Nonlinear regression was applied to estimate the kinetics of the Pressure-Time isotherm curves. The

isotherm was fitted with a sigmoidal model (Hill plot with 3 parameters) using GraphPad Prism

7.03 software (USA). The Hill plot model explained approximately 99% of the pressure variability

in time for all individual data and is described by the following formula:

𝑓(𝑥) = (k∗mn)(onpmn)

Equation 2.4

, where a is the maximum difference in surface pressure reached at the plateau (ΔΠmax), b is the

hill coefficient of sigmoidicity (hill slope at its midpoint) and c is the time for which 50% of

maximum pressure is obtained (t50%).

2.3.7 Ultra Violet Spectroscopy

The UV spectrum of PPA148 was recorded on a Lamda 2 spectrophotometer (Perkin Elmer, UK), at 25 ˚C using a quartz cuvette (Hellma 114-QS) with a pathlength of 1 cm and volume

capacity of 1400 μL. The absorbance was recorded within a wavelength range of 200-550 nm

with spectral slit width of 2 nm and scan speed of 60 nm/min. The drug concentration was 20

μg/mL and it was measured in DI water, ethanol and ethanol/water (80:20).

The linearity of this method was determined by analysing the solutions of PA148 in the range of 2-200 μg/mL. For quantification purposes, the Limit of Detection (LOD) and Quantification (LOQ)

were established by regression analysis of the linear part (calibration curve). Both values were

calculated using the standard error mean (se) and slope (S) taken from the regression line. The

linear regression assumes that the errors between observed and predicted values, i.e. the

residuals of the regression, should not fall outside the straight line (normal distribution) and its


40

variance should be equally distributed (homoskedasticity) (Appendix B, Figure B 4). Otherwise,

the residuals are not constant and normally distributed, leading to false estimations. All the

formulas are as follows:

𝐿𝑂𝐷 = s.s×tuv

Equation 2.5

𝐿𝑂𝑄 = Vw×tuv

Equation 2.6

2.3.8 Turbidimetric Assay

Turbidimetric assay is a preliminary screening test to rapidly evaluate the enhancement of water

solubility of compounds at their early stage of development. Initially, a stock solution of PPA148

(5 mM) in DCM was used to produce a range of concentrations (0, 5, 10, 15, 20, 25, 30, 40, 50,

60, 70, 100, 130 and 160 μg/ml) in water or aqueous buffer. DCM was evaporated before the

addition of the water to a final volume of 4 mL. The samples were left under mild stirring (290 rpm) for 7 days and changes in the light scattering of the samples were tracked using two

techniques. Photon correlation spectroscopy (PCS), also known as dynamic light scattering

(DLS), was the first technique used at 25 °C. A Malvern Zetasizer, Nano-ZS (Malvern

Instruments, UK) was used with a laser of wavelength 623.8 nm and backscatter detection angle

of 173°. The value of the viscosity of water (as the dispersant) and refractive index were used,

respectively 0.889 cP and 1.33211. The samples were placed in disposable semi-micro dynamic

light scattering cuvettes with volume capacity of 1.5 mL (VWR International, UK). The linear

derived count rate (DCR) was recorded and plotted against drug concentration.

UV/Vis spectroscopy was also used as an additional technique to measure the scattered light at

a wavelength of 620 nm at 25 ˚C. Solutions were placed a quartz cuvettes (Hellma 114-QS) with

a pathlength of 1 cm and measured on a spectrophotometer (Perkin Elmer, UK) with slit width

set at 2 nm. The absorbance value at 620 nm was recorded and plotted as a function of drug

concentration.

In the absence of aggregates, scattering from the samples is nearly zero; the onset of

aggregation is denoted by a break in the slope, with scattering (or turbidity) increasing linearly

with concentration. The intersection of the two linear profiles is used as an estimation of the

aggregation concentration.

2.3.9 Water Solubility Test

An excess amount of PPA148 (1 mg) was added to deionized water and 20 mM HEPES, 2 mM

CaCl2, 145 mM NaCl (pH =7.2) buffer (1 mL). The samples were left under mild stirring (290 rpm)

for 7 days. After incubation, samples were centrifuged (5 min at 10,000 rpm) and the supernatant

was freeze-dried overnight and resuspended in ethanol: water (4:1). The saturated concentration


41

was assayed by UV/Vis spectroscopy using a calibration curve as described in the corresponding

methods section (2.3.7).

2.3.10 Quantification of drug:cyclodextrin binding constant by fluorescence

spectroscopy

The increase in aqueous solubility of the antimicrobial agents (PPA148 and rifampicin) obtained

by their incorporation into cyclodextrins was measured by fluorescence spectroscopy. Two types

of modified -βcyclodextrins were used: HPβCD and RAMEB and the binding constant was

measured as reported by Valero et al. (151). Specific volumes of either PPA148 in DCM, or

rifampicin in ethanol were transferred into separate vials and the solvent evaporated. CD

solutions (0, 1, 2, 3, 4, 5, 6, 10, 15, 20, 25 and 30%) in HEPES (20 mM) buffered saline, pH 7.2,

or 50 mM Tris buffer pH 9 were then added to achieve a final drug concentration of 10 µg/mL. The samples were left under mild stirring overnight. The fluorescence intensity of the drug was

measured as a function of concentration. The fluorescence intensity of the drug increases upon

complexation with CDs until the drug is fully complexed with the CD and a plateau is reached.

The fluorescence intensity from PPA148 was measured with a Luminescence spectrometer LS

50 B (Perkin Elmer, UK), and was conducted with an excitation wavelength of 300 nm, and an emission wavelength scan range of 340-530 nm, with emission/excitation slit widths of 5 nm.

Shifts in the fluorescence drug peak at 422 nm for PPA148 and the ratio of 338/419 nm for

rifampicin are due to inclusion complex formation. Assuming that the drug forms a 1:1 complex

with CD, the binding constant (K) was determined by fitting the data with the following equation

(152):

𝐼 = yzpy{f[ab]Vpf[ab]

Equation 2.7

, where 𝐼w, 𝐼 and 𝐼L are the fluorescence intensity from the drug in the absence of cyclodextrin,

at intermediate and infinite cyclodextrin (excess) concentration, K is the binding constant and

[CD] is the molar concentration of free cyclodextrin. A non-linear least squares method was used

to fit the experimental results to Equation 2.7. The differences between the calculated solubility

and the experimentally derived data were minimized by varying the values of the binding constant

(the iterative approach of Nelder–Mead) using the Microsoft Excel Solver function (153).

2.3.11 Phase solubility diagram

Complex formation between cyclodextrin (solubilizer) and drug was also investigated by studying

phase solubility (147). Twelve vials containing an excess of drug, i.e. 1 mg (1.4 mM) for PPA148

and 10 mg (12 mM) for rifampicin, were prepared. The aqueous solution used for PPA148 was

HEPES buffered saline including CaCl2 at pH 7.2; for rifampicin, Tris buffer at pH 9 was used.

Each vial contained varying amounts of cyclodextrin (0, 0.2, 0.4, 0.6, 0.8, 1, 1.5, 2, 4, 6, 10 and


42

15% w/v). The mixtures were left under stirring for 7 days at room temperature (23 °C). For

rifampicin, a photosensitive drug, the vials were covered with aluminum foil to avoid drug

degradation. The undissolved drug was separated from the mixture by centrifugation (Heraeus™ Pico™ 21 Microcentrifuge, UK) at 5000 rpm for 5 min. The supernatant was freeze-dried (ALPHA

1-2 LD plus freeze drier with chemistry hybrid pump RC6). PPA148 was resuspended in

EtOH:H2O (80:20), rifampicin in Tris buffer pH=9 and assayed by fluorescence spectroscopy

(Luminescence spectrometer, LS 50 B, Perkin Elmer, UK). Each sample was scanned over the

range 340-550nm, the excitation wavelength was set at 300 nm, and the scan speed at 120

nm/min. The measurements were performed at 25 °C and the final spectrum was an average of

10 scans. Excitation and emission slit widths were set to 5 mm. A calibration curve for each drug was prepared in the same solvent and measured under the same instrument settings for each

drug on the same day. The molar drug concentration, which was determined from the calibration

curve, was plotted against the CD molar concentration.

The phase solubility diagram is classified into “A” and “B” type depending on the type of complex

formed. Type “A” curves indicate the formation of soluble complex, while “B” types represent the formation of insoluble complexes (147,154). Type “A” curve is obtained when the solubility of a

drug increases with increasing ligand concentration. Negative or positive deviation of linearity

produce the “AL” and “AP” or “AN” curve, “BS” describes complex of limited solubility and “BI”

illustrates the formation of insoluble complex (Figure 2.4) (154).

Figure 2.4: Types of phase solubility diagrams (“A” and “B”) and their deviations (“AP”, “AN”, “BS” and “BI”) (154).

In the “A” type curve, the binding constant (𝐾F), assuming a 1:1 complex and the overall complex

stoichiometry is estimated by using the linear region of the diagram, which is fitted with a linear

model. The slope of the linear regression, the y-intercept representing the drug solubility in the

absence of cyclodextrin (𝑆HIL, the intrinsic solubility) and the solubility in the presence of

cyclodextrin (𝑆w) is used in the following equations:


43

𝐾F =t�J�u

vTU{(V�t�J�u) Equation 2.8

𝑔𝑢𝑒𝑠𝑡: 𝐶𝐷 = 1: Vpvz(V�vz)

Equation 2.9

Due to poor water solubility of hydrophobic drugs, the y-intercept is not always equal to the

intrinsic solubility because of the presence of dimers, trimers or other water-soluble oligomers,

which render the formation of the complexes (147). Therefore, complexation efficiency (CE) is a

more accurate method to determine the solubilizing effect of cyclodextrins. CE is the

concentration ratio between cyclodextrin in the complex and its free form in the system (155). CE

is calculated based on the slope of the linear part of the phase solubility diagram.

𝐶𝐸 = 𝑆w × 𝐾F =t�J�uV�t�J�u

Equation 2.10

The Gibbs free energy of transfer ∆𝐺LGJ is the energy needed for the drug to be transferred from

the aqueous environment to the cyclodextrin cavity and is calculated using the function:

∆𝐺LGJ = −𝑅𝑇𝑙𝑜𝑔(vTU{vz) Equation 2.11

, where R is the gas constant (8.314 JK-1mol-1), T is temperature in Kelvin, 𝑆HIL and 𝑆w are the

molar solubility of the drug in the absence (y-intercept of the linear part of the phase solubility

diagram or intrinsic solubility) and presence of CD, respectively.

2.3.12 Continuous method variation (Job’s plot) (156)

The molecular association of the drug and CD was also studied by NMR spectroscopy. Equimolar

solutions (1mM) of drug and CD in D2O were prepared and mixed in varying volume fractions

while keeping the total concentration constant (1 mM). DIMEB was used instead of RAMEB as it provides a high resolution spectrum from which all protons can be assigned, unlike RAMEB and

HPβCD, which are randomly substituted and therefore lead to spectra with broad peaks (157).

Rifampicin and DIMEB solutions were mixed in volume fractions ranging from 0 to 1 mL with an

increment of 0.1. The inclusion of rifampicin into the DIMEB cavity is evidenced by changes in

the chemical shift observed with selected protons of the cyclodextrin. The spectral peaks of

cyclodextrin hydrogens were assigned according to the literature (158). The settings were the

same as those described in the NMR method section (2.3.4). The resonances of the DIMEB interior hydrogens (in positions 3 and 5) and exterior (position 1) were selected as changes in

those hydrogens indicate inclusion of drug into the cyclodextrin cavity (159). The chemical shifts

of H1 was presented as a doublet at 5.19 and 5.18 ppm, that of H3 was observed as a triplet at

3.90, 3.92 and 3.94 ppm and of H5 as a doublet at 3.84 and 3.86 ppm. The stoichiometry of the

complex was obtained by normalizing the variations of the chemical shifts of the host by its mole

fraction (XCD) and plotting them against cyclodextrin mole fraction (XCD) (160). The shift of CD’s


44

hydrogen resonance depends on the mole fraction of pure drug, pure CD and drug/CD

complexes. A 1:1 complex presents a maximum signal at 0.5 CD mole fraction.

2.4 Results

2.4.1 Synthesis

PPA148 is a novel modified biaryl PBD hybrid analogue and its synthesis is based on a

convergent strategy. The PBD tricyclic scaffold and the lateral polyamidic tail were synthesized

separately and the two pure compounds (#9 and #11) were coupled to give the final PPA148

(#12). The synthesis of PPA148 was performed according to Scheme 2.1, starting from the

intermediate core compound #6 which was previously synthesized by another group member.

The complete synthetic route can be found in Appendix B, which is the one described in the

patent (107). Each reaction was monitored with TLC and upon completion the compounds were characterized with LCMS, both 1H and 13C NMR, TLC and FTIR to verify their structure and the

mass.

The PBD core followed a retrosynthetic pathway. The pro-imine bond was formed on N10C11

via intramolecular oxidative cyclization of the protected amino-alcohol compound #6. Compound

#7 was obtained with a 74.6% yield and 76.2% relative purity (RP). In order to control the final bonding site of the core and tail coupling, the -OH group of compound #7 was protected (93%

yield and 71.2% RP) prior to the saponification reaction (83% yield, 85.9% RP) of the terminal

methyl ester group. Proton NMR revealed all the differences in magnetic environment during the

series of reactions (Figure 2.5, Figure 2.6).

Compound #6 was the starting material for this project. Its 1HNMR generated a signal at 8.73 ppm which was assigned to the amide hydrogen (Figure 2.5). This signal disappeared in

compound #7 which indicated that the pro-imine bond had formed. All proton NMR spectra

contained several overlapping signals due to the similar magnetic environment of the

hydrocarbon chains. The tetrahydropyran protective group added more hydrocarbon chains to

compound #8 which increased dramatically the signal of the protons at 1.63 ppm. The two sharp

peaks in compounds #6, 7 and 8 were assigned to the two methyl groups (Figure 2.5). Although

compound #6 generated the signals at 3.68 and 3.89 ppm, there was a slight shift in signal

towards higher frequencies in #7 and 8. Compound #9 generated only one signal for the methyl hydrogens which validates the hydrolysis of the methyl ester to create the equivalent acid (83%

yield and 85.9% RP) (Figure 2.5). In all spectra traces of solvent signals might be present or

hidden. 13C NMR was performed for all molecules and it showed all the different carbon

environments (Appendix B, Figure B 2).

The tail consisted of three constitutive units at the C8 position which were sequentially connected via an amide bond (Scheme 2.1). The Boc-protected pyrrole-2-carboxylic acid (first unit) was

bonded with the benzofuran carboxylate (second unit) to give the intermediate compound #10


45

which was derivatized with the cyclic secondary amine, i.e. thiomorpholine (third unit), after being

hydrolysed.

The series of amide couplings for the tail formation were successful (Figure 2.6). Compound #10

was formed with a 65.4% yield and 99.4% purity. The final tail was obtained with 56.3% yield and

89.4% purity. The signals at 4.39-4.34 ppm (quadruplet) and 1.36-1.32 (triplet) assigned to the

ethyl group in compound #10 disappeared while the thiomorpholine hydrocarbon chains

appeared at both higher and lower frequencies depending on the neighboring atom (3.93, 3.88

and 2.72 ppm (Figure 2.6). This was an indication that there was no starting material left because if both compounds were present all peaks would be observed in the final spectrum. The four

methyl groups generated from the Boc-protection and the tertiary amine were assigned in the

final spectrum, too, at 1.46 and 3.82 ppm respectively. This was another indication that the

coupling of compound #10 and thiomorpholine succeeded.

Figure 2.5: 1NMR of sequential reaction monitoring for the synthesis of the final PBD core compound (#9).


46

Figure 2.6: 1NMR of sequential reaction monitoring for the synthesis of the final tail compound (#11).

The conversion of compound #10 to the intermediate acid is a crucial step. Citric acid was used

for precipitation of the acid which was formed during hydrolysis with methanol and dioxane.

However, if traces of citric acid remain in the precipitant, it prevents the final amide coupling. The intermediate product with molecular weight of 555 g/mL was formed but it did not interact with

thiomorpholine to give the expected tail product (Scheme 2.2). An attempt to overcome this issue

was to increase the concentration of reagents. Unfortunately, that was not the case, even after

doubling the amount of EDCI and DMAP.

Scheme 2.2: A detailed scheme presenting the second amide coupling of the tail synthesis. When citric acid is present, the reactions did not go to completion and instead only the intermediate product was formed.

The final amide coupling of the PBD core (#9) and tail (#11), de-protection and purification were

successful with 82% yield and 78.1% RP. The proton NMR of PPA148 contained overlapping

signals for the secondary amides, giving a “W” shaped peak instead of two singlets (Figure 2.7).


47

For the hydrogens being involved in double bonds, 9 different signals were generated for each

different environment (Figure 2.7).

In Figure 2.8, there is evidence of the rest of the structure. The hydrogens in the alkyl

hydrocarbon chain were generated at 3.92, 2.72 and 2.04 ppm which agrees with the equivalent

hydrogens in compound #9. The signal of the methyl moieties overlapped as it may be seen in

Figure 2.8. Thiomorpholine group was present at 2.66 and 3.33 ppm because there are 2 different

magnetic environments. The signal of pyrrole of the PBD tricyclic moiety was generated at 2.33

ppm as a multiplet. There are several similar magnetic environments but the signals were not strong. 13C NMR was performed for all molecules and it showed all the different carbon

environments (Appendix B, Figure B 3 and Figure B 4).

Figure 2.7: 1NMR signals for the secondary amide and double bond hydrogens of the final de-protected compound (#12), PPA148.

Figure 2.8: 1H-NMR signals for the hydrocarbon chain, methyl moieties, thiomorpholine and the pyrrole group of the PBD tricyclic core of the final compound, PPA148.


48

Additional information on the structure and functional groups of each compound were collected

by FTIR. The product of each reaction was analyzed as a tool for monitoring each step of the

synthetic route.Figure 2.9, Figure 2.10 and Figure 2.11 outline the changes in functional groups of the PBD core (#9), tail (de-protected #11) and the final compound (#12, PPA148) as they

progress through every synthetic step. This technique reveals the hydroxyl groups which are

difficult to observe in the NMR spectra.

FTIR confirmed the functional groups within the structures of each compound (Figure 2.9, Figure

2.10, Figure 2.11). Particularly, in compound #6 the band at 3328 cm-1 is associated with the -OH and NH stretching vibration bands, which overlap in the spectrum (Figure 2.9). The bands

between 2900-1860 cm-1 result from the asymmetric and symmetric stretching vibration bands

of -CH, -CH2, -CH3 and the band at 1726 cm-1 is generated by the carbonyl group of the ester.

Amide I absorption originates from the C=O stretching of the amide group (coupled to in-phase

bending of the N–H bond and stretching of the C–N bond), which gives rise to bands in the region

between 1660-1560 cm–1. Amide II comes from the N–H bending and C–N stretching vibrations

which generate a band at 1520 cm-1. Amides III is a very complex band resulting from mixtures of several coordinate displacements (-H-N-C bend, -CC stretch and -NH bend). The band at 1263

cm-1 possibly arises from the amide III. The aromatic -C-C- stretches and the C-H deformation

bands of the phenyl ring are generated at 1452-1406 cm-1, 872-768 cm-1 respectively.

Figure 2.9: FTIR monitoring of the core synthesis.

Upon oxidative cyclization, the amide converted into amino-alcohol. Unfortunately, this change was not depicted in the FTIR spectrum because the two peaks were overlapping in #6. However,

the ester, amide I and II bands changed position, which indicated change in the hydrogen

bonding of the molecule (Figure 2.9). This reaction was followed by the protection of the -OH

group of #7 by tetrahydropyran which led to marked changes in the v(OH), amide III v(C-N) and

saturated ether v(C-O-C) region of the spectra. The last step is the hydrolysis of the ester. The

PBD carboxilate (#8) was transformed to the equivalent acid, as indicated by the presence of the

-OH band of the acid.


49

Figure 2.9 shows the spectra of the two-step tail synthesis. The sharp peaks indicate the

formation of crystalline compounds. In compound 11, the absorption band at 1710 cm-1 is

associated with the stretching vibrations of the ester carbonyl group, which disappeared in the spectrum of compound 12 due to the amide coupling with thiomorpholine at that site. The bands

between 1500-1700 cm-1 relates to the -C=O and -H-C-N stretching (amide I), N-H bending, C-

N and C-C stretching (amide II). The bands between 3500-3200 cm-1 result from the NH

stretching vibrations; and that between 3000-2800 cm-1 arise from the CH, CH2 and CH3

stretching.

Figure 2.10: FTIR monitoring the tail synthesis.

The final amide coupling resulted in an amorphous material because the peaks are broad and

not well defined. However, the functional groups were established within the spectrum (Figure

2.11). The final compound is the deprotected #12 which means less -CH, -CH2 and CH3 bonds. This is presented in Figure 2.11 with a change in the band shape at the corresponding region.

Figure 2.11: FTIR monitoring of the final protected and de-protected PPA148.

Table 2.2 presents the mass to charge ratio (m/z) of each compound, estimated by ChemDraw®

and calculated by LC/MS. The experimental m/z was calculated from samples that were

assessed by positive mode ESI. It can be seen that the mass of the parent ion [M+H]+ was as

predicted for all the products. TLC was used to monitor the reactions and become a fast and

straightforward means of identification. The retardation factor (𝑅F) was calculated from the

distance travelled by the solvent and the individual spots created by the compounds (Table 2.2).


50

All compounds travelled up the plate at different rates depending on their affinity with the solvent.

Compound #9 stayed on the baseline because the mobile phase, used (100% ethyl acetate) was

not polar enough to carry the compound with it. That was a way of observing whether the hydrolysis had been completed since compound #8 was able to travel. It was observed that

PPA148 decomposed in methanol and at temperatures above 65 °C (data not shown), by

changing color from light yellow to orange which was confirmed by LC/MS.

Table 2.2: Chemical characteristics of each intermediate and the final compound.

Compound # Estimated M/Z Measured M/Z by LCMS Rf (cm)

6 (C22H30N2O8)

450.4 451.2 0.34

7 (C22H28N2O8)

448.4 449.2 0.38

8 (C27H36N2O9)

532.2 532.6 0.45

9 (C26H34N2O9)

518.5 519.2 N/A

10 (C22H27N3O6)

429.2 429.1 0.44

11 (C24H30N4O5S) 486.2 485.2 0.75

12 (C36H38N6O7S) 698.2 699.3 0.23

2.4.2 Physicochemical characteristics

2.4.2.1 Adsorption at the air/water interface

PPA148’s water solubility (logS = -7.461) and logP (0.76) estimated by ChemDraw suggest a

very hydrophobic drug with low water solubility (Table 1.5). The adsorption of PPA148, rifampicin,

gentamicin sulfate and chlorhexidine digluconate (CHD) at the clean air/water interface were measured in order to present the hydrophilic and/or hydrophobic character of each antimicrobial

agent and to use it as a control in the interaction studies (Section 4.3.3; Section 4.3.4; Figure

4.6; Figure 4.7; Table 4.2) with model lipid membranes presented in Chapter 4. The concentration

of the drug in the small trough was 2 μg/mL. Changes in surface pressure due to the adsorption

of the molecules at the air/water interface were recorded. Figure 2.11 shows the changes in

surface pressure over time after injection of PPA148.

PPA148 reached a maximum surface pressure of 12.34 ± 0.77 mN/m with a Hill slope of 1.03 ± 0.22 (Figure 2.12). The time needed to reach half of the adsorption equilibrium surface pressure


51

was 0.33 ± 0.15 h, which reflects a rather slow kinetic profile. This was probably a result of the

co-solvent (0.5% of DMSO) in the small trough since PPA148 is barely soluble in water and thus

DMSO was used to prepare the stock solution and avoid drug precipitation in water. The data for gentamicin sulphate, rifampicin and chlorhexidine digluconate are not shown here because no

increase in surface pressure was observed. All three drugs are soluble in water at the

concentration used and they did not adsorb at the air/liquid interface.

Figure 2.12: A representative adsorption profile of PPA148 at the air/liquid interface at 23 °C. The drug was injected below a 0.9% NaCl subphase and changes in surface pressure were recorded until a plateau was reached. The representative curve was fitted (red line) based on a 3-parameter Hill equation (Equation 2.4), R2=0.99 (P≤0.0001). The mean and std of maximum surface pressure (12.34 ± 0.77 mN/m), Hill slope (1.03 ± 0.22) and t50% (0.33 ± 0.15 h) was calculated from a set of 3 runs.

2.4.2.2 Solubility and UV profile of PPA148

The UV absorbance spectrum of PPA148 was measured in ethanol/water mixtures (0, 80 and

100% ethanol) over a range of drug concentrations. The experimental data were fitted by linear

regression and met the assumptions regarding the normal distribution and homoscedasticity of

data as described in section 2.3.7 (Appendix B, Figure B 4). The absorbance intensity of the drug

in water was lower than in ethanol, which, in turn, was slightly lower than in ethanol/water (80:20) (Figure 2.13). The shape of the peak was broader in water and better separated in the other two

solvents (Figure 2.13). Figure 2.14A presents the effect of increasing concentration on the

linearity of the UV/Vis method and thus in the Beer-Lambert’s law. For concentrations between

1-15 µg/mL, the drug absorbance in water is proportional to concentration. However, it deviates

at higher concentrations, which is an indication of the presence of molecular aggregates. Visual

observation of the samples showed cloudiness and turbidity for 40-70 µg/mL and drug

precipitation was pronounced for 100-200 µg/mL (Figure 2.15). In ethanol and ethanol/water, the

linearity was extended up to 30 µg/mL, at which point the absorbance reached the highest value regarding the sensitivity of the instrument Figure 2.14A.


52

Figure 2.13: Representative UV/Vis spectra of PPA148 in three solvents: EtOH, H2O, EtOH/H2O (80:10) at 25 °C.

The first derivative of the spectroscopic spectra was used to enhance the separation of

overlapping absorbance peaks at around 290 and 255 nm (Figure 2.14B, C, D). It was observed

that the peak at around 290 nm shifts towards higher wavelength with increasing drug

concentration, whereas the one at 255 nm remains stable. This bathochromic shift occurs at

stages when PPA148 is in water and presents a hyperbolic shape when in ethanol and

ethanol:water (80:20).

UV/Vis spectroscopy was used for the quantification of PPA148. The limits of detection and

quantification in two different systems (100% ethanol and 80:20 ethanol:water) were initially

calculated by a regression analysis of the calibration curve. The assumptions of linear regression

were met and, thus, the results are reliable (Figure 2.14A-inset, Figure B 4). It was found that the

LOD/LOQ (Equation 2.5 and 2.6) for 100% and 80% ethanol were 2.23/6.76 µg/mL and 2.97/8.99

µg/mL, respectively. Once these limits were established, the thermodynamic solubility (𝑆w) was

investigated in water and HEPES buffer (pH=7.2). The experimental S0 was 30.3 ± 1.8 µg/mL

and 60.7 ± 15.6 μg/mL in water and HEPES buffered saline, respectively, after equilibration for

one week.


53

Figure 2.14: (A) Effect of solvent and drug concentration on the absorbance profile. Inset: concentration range where the Beer-Lambert law is followed (𝑦 = 0.04𝑥 + 0.02, 𝑅@ = 0.98 for EtOH/H2O, 𝑦 = 0.04𝑥 + 0.06, 𝑅@ = 0.75 for EtOH, 𝑦 = 0.03𝑥 + 0.05, 𝑅@ = 0.98 for H2O). First derivative of UV/Vis spectrum of a series of PPA148 concentrations in: (B) H2O; (C) EtOH; and (D) EtOH:H2O (80:20). The peaks and troughs of the spectrum profile is presented with zero first derivative. The insets show the changes of λmax with increasing drug concentration.

2.4.2.3 Aggregation and solubility studies

The solubility of PPA148 in aqueous media was further investigated with dynamic light scattering.

DLS is a sensitive technique to detect the presence of small aggregates based on temporal

fluctuations in the intensity of scattered light, as a result of Brownian motion. The derived count

rate (DCR) is a measure of the count rate (average scattering intensity) and is useful to compare

the signal strength in different samples. Figure 2.16 presents the changes in DCR against

increasing drug concentration in water at room temperature (25 °C). The results show that PPA148 generates aggregates at drug concentrations above ~20 μg/mL with an average size of

1µm and high polydispersity index (0.7). Precipitation of the drug was visually observed at

concentration higher than ~30 μg/mL (Figure 2.15).


54

Figure 2.15: Increasing concentration of PPA148 in water 7 days under mild stirring at 25 °C. Visual observation of the precipitation point is at 50 μg/mL.

Cyclodextrins were used as a means of increasing the water solubility of the drugs and avoiding

its self-association. As explained above, the increased scattered light or turbidity reflects the

onset of aggregation (Figure 2.16). In the presence of cyclodextrin (HPbCD), the onset of

aggregation takes place at higher drug concentrations, as seen from a shift in the intersection

point between a flat region and a linear region. The values obtained are 51.1 ± 2.6 µg/mL (PCS)

and 53.9 ± 12.1 µg/mL (UV scattering) (Figure 2.16, Table 2.3).

Figure 2.16: Turbidimetric assay: Effect of cyclodextrin concentration in the aggregation of PPA148 using dynamic light scattering (A) and UV/Vis spectroscopy (B).

Solubility of PPA148 in pure water and HEPES buffered saline pH 7.2 was determined using different techniques. The results are summarized in Table 2.3.

Table 2.3 Solubility values and aggregation concentration of PPA148 in water, HEPES buffered saline pH 7.2 and HPβCD or RAMEB in HEPES obtained by different techniques.

Solvent Solubility (µg/mL) using: Aggregation concentration

(µg/mL) Thermodynamic

solubility Phase solubility

diagram Turbidimetric assay (PCS)

Turbidimetric assay (UV)

H2O 30.3 ± 1.8 - 18.1 ± 4.7 16.3 ± 11.1 HEPES 60.7 ± 15.6 153.8 ± 52.9 - - HPβCD/HEPES - 938.8 51.1 ± 2.6 53.9 ± 12.1 RAMEB/HEPES - 966.8 - -


55

2.4.2.4 Drug/Cyclodextrin inclusion complex formation: binding constant

The formation of an inclusion complex between CD and drug was examined by two methods: binding constant determination by fluorescence spectroscopy (161) (section 2.3.10) and phase

solubility diagram (section 2.3.11) (147). Rifampicin was used as a positive control to evaluate

and establish the method because it is known to form an inclusion complex with CDs (158,162–

164). With the spectroscopic method, the binding of the drug to the CD is reflected by a hyperbolic

function in UV absorbance as the concentration of CD is increased while the total drug

concentration is held constant. The addition of increasing cyclodextrin (either RAMEB or HPβCD)

from 0 to 30 % w/v to a fixed concentration of 10 µg/mL PPA148 and 50 µg/mL rifampicin produces a hyperchromic effect (Figure 2.17, Figure 2.18) reflecting an interaction between the

drug and HPβCD or RAMEB. For both types of cyclodextrin, a good fit (Equation 2.7) of the

experimental data was obtained by assuming a 1:1 stoichiometry. Higher affinity of PPA148 to

RAMEB (102 ± 26 M-1) over HPβCD (63 ± 20 M-1) could be attributed to the different

hydrophobicity of the methylated and hydroxypropyl substitutions. The data were fitted using a

1:2 PPA148:CD complex (data not shown) but the root mean squared error between the

experimental and the fitted data was better when a 1:1 complex was assumed, which illustrated

a better fitting. The binding constants of rifampicin to RAMEB and HPβCD at 50 mM Tris buffer pH 9 were found to be 38 ± 9 M-1 and 22 ± 2 M-1 (Figure 2.18) respectively, which are lower than

values from the literature, namely, 73.4 ± 8.2 M-1 and 68.5 ± 5.2 M-1 for RAMEB and HPβCD,

respectively in sodium tetraborate buffer, pH 9 (162). A much higher binding constant (209.0 M-

1) was found for HPβCD at pH 6 (164) and with DIMEB at PBS buffer pH 5.7 (300 M-1) (158).

Chadha et al. obtained the binding constant using solution calorimetry and the reported values

are 1564 ± 8 M-1 for RAMEB and 1038 ± 11 M-1 for HPβCD in PBS (163). The difference in the

binding constants may be the different methods used and buffer used during the experiment.

Figure 2.17: Changes in the fluorescence intensity of PPA148 with increasing cyclodextrin concentration (HPβCD or RAMEB) at 25 ˚C (A). The lines represent the non-linear regression fit (Equation 2.7) of the experimental data points assuming a 1:1 complex. The phase solubility diagram (B) was built by adding an excess of drug to solutions of cyclodextrin to measure drug


56

solubilisation. In both experiments, the solution used was HEPES buffered saline including CaCl2 at pH 7.2.

Figure 2.18: Changes in the fluorescence intensity of rifampicin in increasing cyclodextrin concentration (A) show the drug’s affinity to HPβCD and RAMEB at 25 ˚C. The lines represent the non-linear regression fit of the experimental data points assuming a 1:1 complex. The phase solubility diagram (B) was built to observe the approximate amount of drug that can be solubilized with CDs and gave an indication of the complex stoichiometry and affinity. In both experiments, the solution used was Tris buffer pH 9.

The phase solubility diagram was constructed by measuring the apparent drug solubility (of

rifampicin and PPA148) as a function of increasing CD concentration. It was used as a tool to

identify the drug behavior and its solubility profile based on the shape of the diagram (147). The

phase solubility diagram obtained for rifampicin agrees with published data in terms of shape, binding constant and complex stoichiometry (164). Rifampicin presents an increase in

solubilization as the amount of CD is increased, up to ca. 34 mM (5% w/v) HPβCD (Figure 2.18B).

In the plateau region, the concentration of the drug is 13.6 ± 1.0 mM (11.2 ± 1.2 mg/mL).

Additional quantities of CD do not alter the diagram, suggesting that saturation has been

reached. The intercept of the phase curve with the y-axis (Sint) should reflect the intrinsic drug

solubility. Here a value of 4.9 ± 0.4 mM (4.0 ± 0.3 mg/mL) was obtained, which is higher than the

water solubility given by the manufacturer (2.5 mg/mL or 3 mM). The linear region of the diagram

suggests the formation of a 1:1 complex with a binding constant of 349 ± 75 M-1, which is higher

than the value found using the fluorescence spectroscopy method (Figure 2.18A). The ratio of cyclodextrin concentration in complex over free form (CE) found from the phase solubility

diagram is CE = 1.7 ± 0.4 which is higher than that found by Tewes et al. (CE = 0.3) (162). Similar

solubility behaviour was observed for PPA148 with both RAMEB and HPβCD (Figure 2.17B).

However, the phase solubility test was performed only once because of the large quantities of

drug needed to perform the experiment, thus, only qualitative results are presented. PPA148 (1 mg) was completely solubilized (0.99 mg/mL or 1.4 mM) with 6 mM (0.8%) RAMEB and 15 mM

(2.0%) HPβCD. In the solubility diagram with RAMEB, the intercept of the linear part gives 𝑆HIL =


57

153.8 µg/mL (0.2 mM), whereas in that with HPβCD to 191.3 µg/mL (0.3 mM) (Table 2.3). As in

the case of rifampicin, the solubility obtained from the phase solubility diagram is around double

the thermodynamic concentration.

Table 2.4: Binding constant of PPA148 and rifampicin as obtained by two different techniques.

CD

Binding Constant (M-1)

PPA148 Rifampicin Rifampicin

(reported values)

Binding assay

Phase solubility diagram

Binding assay



HPβCD 63 ± 20 71.2 22 ± 2 349 ± 74 18* 120/125** 68.5 ± 5.2***

RAMEB 102 ± 26 161.1 38 ± 9 - - - 73.4 ± 8.2*** *The buffer used in the experiment was 10 mM Tris pH 6.9 and ionic strength of 10 mM. ** The buffer used in the experiment was 30 mM citrate/10 mM Tris pH 6.9 with ionic strength of 0.8 mM. *** The buffer used in the experiment was 50 mM sodium tetraborate pH 9.

Gibbs free energy change (∆𝐺��w ) values for the process of transferring a drug from pure water

to aqueous cyclodextrin solution were calculated based on Equation 2.11 and presented in Table

2.3. ∆𝐺��w values were found to be negative at all concentrations of CD in all cases, indicating

the spontaneous nature of complex formation. The values were negative upon increasing CD

concentration, indicative of more favorable solubilization reaction as the concentration of CD

increased.

Table 2.5: Gibbs free energy of transfer of rifampicin and PPA148 from pure water to aqueous CD solutions at room temperature calculated using Equation 2.11.

HPβCD or RAMEB (%)

∆𝑮𝑻𝑹𝟎 RIF: HPβCD

∆𝑮𝑻𝑹𝟎 PPA148: HPβCD

∆𝑮𝑻𝑹𝟎 PPA148: RAMEB

0.2 N/A -67 N/A

0.4 N/A N/A -491

0.5 -443 N/A N/A

0.6 N/A -29 -700

1 -627 -420 N/A

1.5 -860 N/A -704

2 -1066 -690 N/A

4 N/A -480 -562

5 -1091 N/A N/A

10 -1160 -672 -672

15 -1131 -691 -685


58

2.4.2.5 Stoichiometry of drug/CD complexes

The stoichiometry of the complex was examined using the Job’s plot method from NMR data (section 2.3.12, Figure 2.19). The normalized NMR signal is plotted against the mole fraction of

one reactant (molar fraction of cyclodextrin in this case), while the total reactant concentration is

kept constant (1 mM). Figure 2.19 presents changes in protons at position 1 (external) and 5

(internal) in a series of cyclodextrin molar fraction for rifampicin and PPA148, respectively. The

Job’s plot of H1 for rifampicin and DIMEB reflects the formation of a 1:1 inclusion complex, with

a maximum at XCD=0.5. This stoichiometry is in agreement with published data (158).

Figure 2.19: Job’s plot for rifampicin/DIMEB (A) and PPA148/DIMEB (B) complexes. In the case of rifampicin, the normalized chemical shift (Δδ*XCD) of the H1 protons of DIMEB were plotted against the molar fraction of DIMEB (XCD). For, PPA148, protons at position 5 of the cyclodextrin structure.

In order to build a Job’s plot, PPA148 was used at very low concentration (20 µg/mL, or 0.03

mM) because of its poor solubility in water (Table 2.3). The spectra were noisier, due to the low

concentration, but the chemical shift of H5 shows a skewed Job’s plot with a maximum around

XCD=0.6, thus revealing a dominating 1:2 PPA148/DIMEB complex, namely, the drug is

encapsulated by two cyclodextrins in two different positions.

2.5 Discussion

A novel biaryl pyrrolobenzodiazepine (PBD) hybrid analogue, PPA148, was successfully

synthesized based on the reported synthetic route (107). Specifically, a convergent synthetic

strategy was used, based on the separate synthesis of the PBD tricyclic scaffold (core) and the

lateral polyamidic tail. The two constituents of the final core and tail molecules were coupled to

give the final PPA148 with 82% yield and 78.1% relative purity. Based on the ICH guidelines, the

purity of new compounds should be calculated based on an analytical standard (165). Here, the

purity was calculated based on PPA148’s peak areas as a percentage of all other peak areas in the LC/MS chromatograms because there was no analytical standard. The drawback of this

approach is that any non-UV active impurities present in the sample are not detected.


59

Affinity for the air/water interface, a measure of amphiphilicity, was investigated at the molecular

level using the Langmuir trough technique (Figure 2.12). PPA148 adsorbed at the air-water

interface and reached a surface pressure of 12.3 ± 0.7 mN/m. It was expected that the adsorption

would be instantaneous because of the properties estimated by ChemDraw (logP = 0.76 and

logS = -7.461). However, PPA148 reached half of the maximum pressure within 20 min and from

that point it took another 2 h to reach the final maximum. This slow kinetic profile can be explained

by two possible transport-limiting steps. The drug was injected below the air/water interface

under mild stirring in order to mix the drug with the subphase. The initial adsorption was delayed due to the diffusion of the co-solvent (0.5% DMSO) in the aqueous subphase of the trough.

DMSO is miscible with water and forms strong hydrogen bonding interactions with 3 water

molecules (166). It was introduced into the aqueous system as a solvent for the drug via injection

which possibly affected the kinetic profile of the drug absorption process (167). The second stage

was due to the saturated surface after 40 min (Figure 2.11). In the early stages of adsorption, i.e.

after drug injection, the surface is relatively empty and molecules adhere as fast as they reach

the surface. However, as the surface gets saturated by drug molecules, the diffusion at the

interface might be slowed down before the pseudo-steady equilibrium is reached (168).

PPA148 is a hydrophobic antibiotic with low water solubility. It presented a water solubility of 30.3

± 1.8 µg/mL and aggregation occurred at concentrations above 18 µg/mL (Table 2.3).

Sedimentation arose by attraction forces between drug molecules and repulsion forces between

drug and water, which are responsible for low-concentration agglomeration, possibly leading to

drug sedimentation at around 50 μg/mL (Figure 2.15). (169). In addition, the fact that the

absorption intensity of PPA148 decreased with increasing water content in the EtOH/H2O

mixture (Figure 2.13) and in 100% water deviated from linearity with increasing amount of drug

(Figure 2.14A and Figure 2.16) indicated the formation of agglomerates/aggregates. The

thermodynamic water solubility (30.3 ± 1.8 µg/mL) was higher than the aggregation concentration

(Table 2.3), which implies the presence of both monomers and small aggregates (169). The

aggregates were not optically visible at 30 µg/mL which implies that the aggregates might have

been dimers, trimers or other water-soluble oligomers whose size is below the visible light

wavelength range (400-700 nm) (110,155).

Drug aggregation can have a negative impact on antimicrobials, because their bioavailability is

reduced and less substance is available to interact with microorganisms (170). Cyclodextrins

were successfully used to improve the solubility of the drug through inclusion complex formation.

RAMEB and HPβCD were shown to solubilize PPA148 up to approximately 50 µg/mL in HEPES

buffered saline pH 7.2 (Table 2.3) in HEPES buffered saline (pH 7.2) by delaying the formation

of aggregates. Rifampicin and PPA148 possess hydrophobic moieties (Figure 1.9, Figure 2.2)

capable of entering the apolar CD cavity through inclusion complex formation (171). One of the aims of this project was to avoid the formation of aggregates and improve drug water solubility

by forming cyclodextrin inclusion complexes, as a way to improve bioavailability and as a first


60

step in the formulation, which would then involve encapsulation in liposomes (Chapter 3).

Rifampicin, an antimicrobial agent that would benefit from an optimized formulation, was used as

a control drug because it is hydrophobic and is known to form complexes with cyclodextrins (162). The data obtained from fluorescence spectroscopy was fitted by a 1:1 binding constant

both for rifampicin and PPA148 with RAMEB and HPβCD. Based on the calculated binding

constant for a 1:1 complex from the fluorescence intensity, PPA148 present a higher affinity for

RAMEB (102 ± 26 M-1) than HPβCD (63.3 ± 20.3 M-1) (Figure 2.17A, Table 2.4). The method

based on phase solubility showed a lower value for the PPA148/RAMEB complex (161.1 M-1) but a similar value for PPA148/HPβCD (71.2 M-1) (Figure 2.17B, Table 2.4). In the spectroscopic

method, the total drug concentration is kept constant at a low value but the free (soluble) drug

decreases due to inclusion into the CD cavity, up to a plateau value. Instead, in the phase

solubility diagram, an excess of drug is added, which is solubilized by increasing amounts of

cyclodextrin (73). In contrast to the binding constant and phase solubility methods, NMR (Job’s

plot) showed that a 2:1 PPA148/DIMEB complex was formed. Both the phase solubility diagram

and fluorescence spectroscopy illustrated higher affinity for both drugs towards the randomly

methylated cyclodextrin (RAMEB). This is also observed for rifampicin (162,163) and hemisuccinate ester, prodrug of Δ9-Tetrahydrocannabinol, (172). The methylation of βCD

creates a more hydrophobic environment and enhances molecular flexibility, which allows an

increased adaptability of the cyclodextrin towards the guest molecule (163).

Although the findings on the complex stoichiometry for rifampicin agree with the literature, the

binding constant values differ slightly from previous reports (158,162,164). Rifampicin has three ionization states (Figure 1.9) with a maximum water solubility at pH between 6-9 because the

predominant state is anionic. In this study, the binding constant between rifampicin and

cyclodextrin was determined at pH = 9 using two methods, a phase solubility diagram and the

shift in fluorescence intensity, with slightly different results (Table 2.4). The binding constants for

rifampicin with RAMEB or HPβCD reported in the literature vary between 500 and 5000 M-1

depending on the pH, temperature, buffer solution and calculation method. The binding constant

K for a 1:1 rifampicin/HPβCD complex calculated from the phase solubility study was 349.4 ±

74.8 M-1 and it approaches the value reported by He et al. (164) (227.3 M-1). However, the binding

constant determination by fluorescence spectroscopy (assuming a 1:1 complex) resulted in lower

K values (22.5 ± 1.7 M-1 for HPβCD and 38.4 ± 8.6 M-1 for RAMEB in Tris buffer pH 9), compared

to those found by the phase solubility diagram under the same experimental conditions, and to

those published by Tewes et al. (68.5 ± 5.2 M-1 for HPβCD and 73.4 ± 8.2 M-1 for RAMEB in

sodium tetraborate pH 9) (162). Overall, the interaction with RAMEB was stronger compared to

HPβCD, which is in agreement with Chadha et al., who found binding constants of 1038 and

1564 M-1 using solution calorimetry (163). Job’s plot actually showed that a 1:1 rifampicin/DIMEB

complex was formed, which confirms the binding constant and phase solubility assumption of the calculations. It has been reported that it is the piperazine moiety of rifampicin that inserts into the


61

CD cavity with both types of CD used in this study (Figure 2.20). This postulated structure of

rifampicin/CD complex is established by using 1H and 13C-NMR and 15N-NMR (173).

Figure 2.20: Schematic representation of rifampicin/CD inclusion complex as it was found by molecular simulation by He et al. and Tewes et al. (162,164).

Another measure of solubilizing efficiency is to determine the molar concentration ratio of

cyclodextrin in a complexed form to the free cyclodextrin; this is referred to as complexation efficiency (CE). According to Loftsson and Brewster (155), CE is a better measure of efficiency

because it is less sensitive to errors related to the estimation of the intrinsic solubility. The

aggregation of highly hydrophobic drugs in aqueous solutions may affect the slope obtained in

the phase solubility diagram and may cause deviations in the value of the y-intercept (𝑆HIL) from

𝑆w, leading to possible inaccuracies and over-estimation of the binding constant (73). The CE

calculated for rifampicin was 1.8 at pH 9 (Figure 2.18B) which is higher than values cited in the literature using the same method at pH 9 (0.347) (174). This means that one in two cyclodextrin

molecules are forming a water-soluble complex with rifampicin, compared to the cited one in four.

The values of CE for PPA148/HPβCD and PPA148/RAMEB were found to be 0.05 and 0.1,

respectively (Figure 2.17B). CE is correlated with drug hydrophobicity and S0 (155). Hydrophobic

drugs with S0 higher than 0.01 mg/mL possess a high CE (175). For example, it has been found

than clotrimazole (𝑆w = 0.037 mg/mL) has a CE of 0.05 with HPβCD in water, hydrocortizone (𝑆w

= 0.4 mg/mL) has a CE of 1.3 with HPβCD at pH 7 and fluoxetine HCl (𝑆w = 12.6 mg/mL) has a

CE of 4.7 at pH 3.6 (175).

The driving force for complex formation is a combination of electrostatic interactions, van der

Waals forces, hydrogen bonding and charge-transfer interactions (146,175,176). The binding

process can be separated into 3 steps (146). The first step is the exclusion of cavity-bound high

energy water molecules into the gas phase, accompanied by changes in CD conformation energy. The second is the solvent (water) rearrangement and transformation of the excluded


62

gaseous water molecules into a liquid phase which is accompanied with a negative enthalpy and

entropy change. The third step is the fit of the hydrophobic guest molecule into the nonpolar CD

cavity accompanied by host-guest intermolecular interactions and a change in the conformation energy of the host (176). Therefore, van der Waals forces and hydrophobic interactions constitute

the basic driving force for complex formation and hydrogen bonding, and electrostatic interaction

can affect the conformation of complexes (146).

The estimated Gibbs free energy of transfer for both rifampicin and PPA148 were negative,

indicating that the complexation is spontaneous, as expected. The interaction of both drugs with RAMEB were found to be sterically favored compared to other CDs, because the methyl

substitution creates a more hydrophobic microenvironment compared to the hydroxypropyl

substitution (163).

2.6 Conclusion

The novel antimicrobial agent was successfully synthesized and characterized. It was found that

PPA148 is barely soluble in water (30.3 ± 1.8 µg/mL) and self-associates when dispersed in an

aqueous environment above its water solubility level. To enhance water solubility and avoid

aggregation, cyclodextrins (RAMEB and HPβCD) were used to form an inclusion complex with

PPA148. Water solubility increased 6-fold through the incorporation of PPA148 into

cyclodextrin’s cavity and a higher affinity was obtained with RAMEB compared to HPβCD.

Enhancing water solubility was the first step towards developing a formulation for PPA148 to

enhance its antibacterial efficacy against Gram negative bacteria. The next two chapters focus

on incorporating the PPA148/RAMEB complex into fluid liposomes (fluidosomes) to improve its antibacterial activity and drug uptake, which are assessed using biophysical approaches.

Liposomaly encapsulated drug/cyclodextrin inclusion complexes

63

Chapter 3 Liposomaly

encapsulated drug/cyclodextrin

inclusion complexes


64

3.1 Introduction

The development of new antimicrobial agents and the understanding of their mechanism of

action is a pressing need because of the widespread proliferation of drug resistance, especially

in Gram-negative bacteria, the most common cause of nosocomial infections in Europe and USA

(177,178). Antimicrobials are losing their effectiveness at an increasing rate, resulting in growing

difficulty in treating bacterial infections. The World Health Organization (WHO) has developed the global priority pathogen list, which includes critical, high and medium risk pathogens for public

health (Table 1.2) (44). Among the critical and high-risk pathogens are Enterococcus faecium,

Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas

aeruginosa and Enterobacter spp, also known as the ESKAPE pathogens, which cause hospital

acquired infections (57). In the same report in 2017 (57), the WHO stated the need to support

R&D, and furthermore, discover and develop new antibiotics against multi- and extensively drug-

resistant Gram negative bacteria.

The objective of this work was to develop a suitable carrier to increase drug uptake and drug

efficacy against Gram negative bacteria. This chapter describes our attempt at developing and

characterizing a formulation for the newly discovered antimicrobial PPA148, combining

cyclodextrins and liposomes, based on their respective characteristics which were considered

advantageous to achieve criteria of improved efficacy and higher solubility. The rationale behind

this formulation is that the liposomal carrier (fluidosomes) will enhance the permeation of PPA148 through the Gram negative OM, while cyclodextrins enhance its water solubility and permeation

through the IM.

Fluidosomes are fluid liposomes composed of DPPC/DMPG (18/1), which are characterized by

their ability to fuse into lipid membranes (179). Beaulac and co-workers were the first to evaluate

the efficacy of tobramycin encapsulated into several types of liposomes against Pseudomonas aeruginosa (102,103). Their results showed antibacterial efficacy of the encapsulated

tobramycin, and that the most efficient carriers were the fluid (fluidosomes) rather than the more

rigid liposomes. Sachatelli et al. then published encouraging results on fluidosomes, using

immunoelectron microscopy and Gram staining techniques to show that fluidosomes may fuse

with the bacterial cell envelope and release tobramycin in the cytoplasm (179). Until now, only a

few antibiotics encapsulated in fluidosomes have been examined for their increased efficacy

(Table 1.4).

Molecules with poor water solubility, such as PPA148, are mainly loaded in the hydrophobic tail

region of the bilayer, leading to limited drug loading and immediate drug release (180). Therefore,

the first step of this work was to increase the water solubility of PPA148 by using

pharmaceutically acceptable βCD derivatives, as we showed in Chapter 2. Water solubility

increased 5-fold by incorporating PPA148 into the cavity of cyclodextrins (Figure 2.16, Table 2.3).


65

In addition to the higher solubility achieved, RAMEB was also chosen because of its expected

permeation through the phospholipid bilayer of the IM (181). RAMEB has been reported to induce

phospholipid exchange (181), which could be of great assistance for the transport of drug through the IM. Phospholipids with longer alcyl chains are difficult to extrude from liposomes because in

complexing with CD, their chain may protrude into the aqueous environment, which is not

energetically favorable (181). There is also a limit on the chain length that can be fully complexed

with β-cyclodextrins: phospholipids with 14 carbons can be fully incorporated into the CD cavity

but not those with 16 carbons (182,183). DPPC (Figure 3.1) was found to interact with CDs less

than other phospholipids, due to its chain length and its high phase transition temperature (Tm at

41°C), thus forming a tightly packed gel phase lipid film at room temperature (181,181).

Figure 3.1: Structure DPPC (16:0) and DMPG (14:0).

In this chapter, PPA148/RAMEB complexes (presented in Chapter 2) were incorporated into

fluidosomes to enhance the antimicrobial efficacy of the novel compound. Empty fluidosomes

were initially used to develop a robust manufacturing method. The thin film hydration method

was used to form the fluidosomes, followed by extrusion for size reduction to ~100 nm.

Fluidosomes encapsulated within the drug/CD complex were separated from unloaded vesicles

using size exclusion chromatography. The elution profile of the lipids and drug through the size

exclusion column was quantified by UV spectroscopy. The aliquot containing most of the loaded fluidosomes was used to determine, spectroscopically, the encapsulation efficiency and loading

capacity of the formulation. A microbiological assay (Kirby Bauer or disk diffusion assay)

(184,185) was carried out to examine the efficacy of PPA148 alone, PPA148/RAMEB complex

and PPA148/RAMEB in fluidosomes. The inhibition zones created by PPA148 (both alone and

formulated) were compared with that of rifampicin (positive control) and gentamicin (negative

control). For this assay, a disk preparation method was developed to generate valid results

comparable to commercially available disks.


66

3.2 Materials

Lipids: 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC, 16:0 PC) and 1,2-dimyristoyl-sn-

glycero-3-phosphoglycerol, sodium salt (DMPG, 14:0 PG) were obtained from Avanti polar lipids

(Alabama, USA) and supplied by Stratech Scientific (Newmarket, UK).

Microbiological Assay: Blank, rifampicin (30 μg) and vancomycin (30 μg) susceptibility disks,

agar powder and Muller Hinton broth for microbiology were purchased from Oxoid (UK). LC/MS

grade water was supplied by Merck, UK. E. coli bacteria DH5α were obtained from Invitrogen

Life Science Technology (Thermo Fisher Scientific, UK).

Salts used for buffer and reagent preparation: Sodium chloride (NaCl), magnesium chloride

(MgCl2), monosodium phosphate (NaH2PO4), sodium hydroxide (NaOH) Ammonium

thiocyanate, ferric chloride and HEPES (>=99.5%) were supplied by Sigma-Aldrich, UK.

Cyclodextrins: Hydroxypropyl-β-cyclodextrin (HPβCD), randomly methylated β-cyclodextrin

(RAMEB) supplied from Sigma-Aldrich, UK.

Solvents: Dichloromethane (DCM), chloroform and deuterated water were purchased by Sigma-

Aldrich, UK. The ultrapure water at 18.2 MΩ was produced by a Purelab Ultra machine from

ELGA process water (Marlow, UK).

3.3 Methods

3.3.1 Formation of Large Multi-Lamellar Fluidosomes

Fluidosomes consisted of DPPC/DMPG in a molar ratio of 18:1 and were manufactured by the thin film hydration method (186). The mixture of lipids (10 mg) was dissolved in 1 mL of

chloroform in a round-bottom flask and sonicated in a sonicating bath (Fisherbrand®, FB11203,

Fisher Scientific, UK) at 25°C for approximately 1 min until achieving a clear solution. The solvent

was evaporated using a rotary evaporator (Rotavapor® RII, Büchi, Switzerland) attached to a

vacuum pump (KNF Lab, UK). The temperature of the water bath was set to 40°C and the rotation

speed to approximately 190 rpm. The resulting thin lipid film was kept under reduced pressure in

a pyrex vacuum desiccator for 12 h to further remove traces of organic solvent. The lipid film was

then solvated by adding 4 mL of phosphate buffer pH 7.4 (50mM PBS, 40mM NaH2PO4 and 26 mM NaCl) or 20 mM HEPES buffered saline pH 7.2 (145 mM NaCl and 10 mM CaCl2) under

vigorous stirring and bath sonication (Fisherbrand®, FB11203, Fisher Scientific, UK) at 40°C to

facilitate vesicle formation. The lipid suspension was dipped into an ice bath (4°C) for 1 min and

then placed in the water bath (40°C) for another minute. This temperature changing sequence

was repeated 5 times followed by a 1 h resting period at 4°C before using them for size reduction.


67

Empty liposomes of the desired size were used to validate the separation method of free

molecules from the drug-in-CD encapsulated liposomes using a PD10 G25 Sephadex column

(GE Healthcare, UK). The method is described in section 3.3.5.

3.3.2 Size reduction methods for the manufacture of Fluidosomes

Size reduction of fluidosomes was carried out by two different methods: probe sonication and

extrusion, to manufacture homogeneous lipid dispersions with unilamellar vesicles. Following the

probe sonication, the lipid suspension was vortexed (Cyclone-Vortex mixer CM-1, Nickel-Electro

Ltd) for 60 sec at 2800 rpm, sonicated (Soniprep 150 Plus ultrasonic disintegrator 120V 60Hz, London, UK) for 5 min with a microprobe at an amplitude of 10 microns and left at room

temperature for one hour prior to further use.

The second method was extrusion, which was carried out by placing 1 mL of suspension in a

tight gas syringe on a mini-extruder (Avanti Polar Lipids) thermostated at 55 °C and manually

extruded through a 0.1 μm polycarbonate membrane 11 times. The extruded liposomes were kept at 4°C and were extruded again under the same experimental set up after 12 hours. The

double extruded liposomes were stored for further use.

The SUVs vesicles formed with the two different methods were compared in terms of

hydrodynamic diameter using photon correlation spectroscopy.

3.3.3 Preparation of drug-cyclodextrin inclusion complexes

A 1:1 complex of drug and RAMEB was used for its encapsulation into liposomes. Drug and

cyclodextrin were mixed in a 1:1 molar ratio in HEPES buffered saline at pH 7.2. The mixture

was left under stirring (190 rpm) for 24 hours.

3.3.4 Incorporation of drug/cyclodextrin complexes into liposomes

For the entrapment of drug/CD complexes into fluidosomes, 20 mg of DPPC/DMPG mixture

(molar ratio of 18:1), 0.7 mM PPA148 and 0.7 mM RAMEB were used. The 1:1 PPA148/RAMEB

complex was prepared as described above in 20 mM HEPES buffered saline, pH 7.2, and was

used to hydrate the lipid film as described in section 3.3.1. Extrusion or sonication was applied

for size reduction of liposomes (section 3.3.2) and the separation of the liposomes from the

unentrapped drug/CD was carried through size exclusion chromatography (section 3.3.5).

3.3.5 Size Exclusion Chromatography

A Sephadex™ G-25 PD-10 column (GE Healthcare, UK) (packed bed dimensions of 1.45 x 5 cm

and particle size range 85-260 μm) was equilibrated with 15 mL (3 washes of 5mL) of 20 mM

HEPES buffered saline, pH 7.2. The flow through the column was driven by gravity. Each

liposome mixture (1 mL) was allowed to penetrate the gel entirely and the eluted volume was


68

discarded. Then, aliquots of HEPES buffered saline (1 mL) were added to the column

sequentially (8 times) to elute various fractions which were collected separately. Each fraction

was characterized for its hydrodynamic diameter, derived count rate, lipid and drug concentration so as to investigate which aliquot contained the majority of the loaded liposomes. Zeta (ζ)-

potential was used to measure the electrokinetic potential properties of liposomes.

3.3.6 Photon Correlation Spectroscopy (PCS)

PCS, also known as Dynamic Light Scattering (DLS), was used to investigate the presence of

particles dispersed in a liquid medium and to determine their average hydrodynamic diameter, zeta-potential and derived count rate. This method measures the scattering of light coming from

the Brownian motion of the suspended particles. The particles’ movement is related to the particle

size, which can be measured using the Stokes–Einstein relationship as follows:

𝑑� =��

sA�b Equation 3.1

where 𝑑� is the hydrodynamic diameter, k is the Boltzman’s constant, T is the absolute

temperature, η is the solvent viscosity and D is Brownian motion diffusion coefficient.

The sample was placed into a disposable semi-micro dynamic light scattering cuvettes (VWR

International, UK) and the size and derived count rate was measured by a Malvern Zetasizer

(Nano-ZS, Malvern Instruments, UK) at 25 ˚C with a laser wavelength of 623.8 nm, backscatter

detection angle of 173°, viscosity of dispersant of 0.889 cP (water) and refractive index 1.590. Placing detectors at 173° enables scattered light signals of low intensity originating from smaller

particles to be revealed and excludes excess scattered light (187).

The Malvern Zetasizer (Nano-ZS, Malvern Instruments, UK) was also used for measuring the ζ-

potential by electrophoretic light scattering spectroscopy. It provides information on the

electrokinetic charge of Fluidosomes and the effect of PBS, pH 7.4 (40 mM NaH2PO4 and 30 mM NaCl) on their stability. A folded disposable capillary cell (Malvern Panalytical, UK) was used to

measure the voltage signal at 25 ˚C with the dielectric constant set at 78.5. The dispersant

refractive index and viscosity were set as per the light scattering experiments. The equilibration

time for both size and zeta-potential measurements was 120 sec and the appropriate attenuator

position was automatically determined by the Nano software. All measurements were carried out

in triplicate.

3.3.7 Stewart assay

The determination of lipid concentration in the final formulation was achieved using the

colorimetric Stewart assay (188). This assay is based on the fact that DPPC forms a complex

with a reagent comprising of ferric chloride hexahydrate (0.17 M) and ammonium thiocyanate


69

(0.4 M) in DI water. The ferrothiocyanate reagent is insoluble in chloroform unless it forms a

complex with DPPC.

A calibration curve was established from standard DPPC solutions in chloroform. All standard

samples were prepared by vigorously mixing the aqueous solution (2mL) containing the reagent

and the chloroform (2mL) containing increasing concentration of lipids (0-0.05 mg/mL). In order

to ensure the full partition of the complex into the organic phase, the mixture was vigorously

mixed and then centrifuged (AllegraTM X-12 centrifuge, Beckman Coulter, UK) for 10 min at

1000 rpm to separate the two phases. The lower layer (organic phase) was collected and its absorbance measured at 467 nm. A linear relationship was observed between the absorbance

and the DPPC concentration in the complex and used as a calibration curve.

For the test samples, the suspension before separation and the 8 eluted aliquots from the PD10

column were used to determine the lipid content in the starting and purified liposomes to detect

any material loss during the process. The samples were lyophilized in an Alpha 1-2 LDplus freeze dryer (Martin Christ Gefriertrocknungsanlagen GmbH, Germany), connected with a Chemistry

hybrid RC 6 pump (Vaccubrand GMBH, Germany) and then re-suspended in chloroform. A

volume of 2 mL was transferred in chloroform (2 mL) which was added into the reagent solution

to form the biphasic system. All tubes were vortexed as the standard samples and the DPPC

concentration of the test tubes was interpolated from the calibration curve. All measurements

were carried out in triplicate.

3.3.8 Drug quantification

Drug quantification was carried out using UV/Vis spectroscopy to determine the encapsulation

efficiency (EE) and the drug loading (DL) of liposomes. Measurements of the LOD and LOQ were

reported in Chapter 2, section 2.3.7. The UV spectrum of PPA148 was recorded on a Lamda 2

spectrophotometer (Perkin Elmer, UK), at 25˚C using a 1 cm pathlength quartz cuvette (Hellma

114-QS). The spectra were recorded within the wavelength range 200-550 nm, with spectral slit

width of 2 nm and scan speed of 60 nm/min. PPA148 was extracted from all 8 aliquots collected from the size exclusion chromatography used to purify the formulation. Each aliquot was washed

with DCM 5 times and the organic phase was collected, dried with magnesium sulfate (MgSO4)

and evaporated using a rotary evaporator (Rotavapor® RII, Büchi®, Switzerland) attached to a

vacuum pump (KNF Lab, UK). The resulting dry sample was re-suspended in ethanol:water

(80:20) and the unknown drug concentration was calculated by interpolating from a linear

calibration curve, which was determined under the same experimental conditions. The area

under the curve (AUC) of all samples was used for drug quantification and calculated using GraphPad Prism 7.03 software (GraphPad software Inc., USA). The number of test and

calibration curve sample measurements was 4 and 3, respectively. The Encapsulation Efficiency

(EE) and Drug Loading (DL) were calculated using the following the equations,


70

𝐸𝐸(%) = a��a�a�

× 100% Equation 3.2

𝐷𝐿(%) = a��a�a�pa��a�

× 100% Equation 3.3

, where 𝐶K is the un-entrapped drug concentration, 𝐶� and 𝐶� are the concentration of lipids and

drug added into the liposome system, respectively.

3.3.9 Kirby Bauer assay

The Kirby Bauer method, also known as the disk diffusion assay, is used for testing the resistance

or susceptibility of bacteria to drugs and chemicals (184,185). It is a well-established assay by

the World Health Organization (WHO) and widely accepted by many clinical laboratories. Escherichia coli (E. coli) was cultured in a Mueller-Hinton (MH) agar (21 g/L MH broth and 17 g/L

agar) plate from a gel-bead for 24 h at 37 ˚C. A few bacterial colonies were taken with a loop

from the cultured disk, placed in MH broth and shaken while incubating for 24 h at 37˚C. The

growth of the bacterial culture was monitored by measuring the optical density (OD) of the

bacterial suspension at 600 nm using a single beam JENWAY 6300 Visible Spectrophotometer

(Staffordshire, UK) with resolution of 1 nm and spectral bandwidth of 8 nm. An OD600 of 1, which

was achieved for this experiment, means that there are ~8´108 cells/mL in the suspension. To

stop bacterial growth, 1 mL of the suspension was centrifuged at 10,000G for 5 min and the

bacterial pellets were re-dispersed in sterile physiological saline (0.9% NaCl). An aliquot of this

suspension (100 μL) was swabbed on an MH agar plate, and filter disks impregnated with

antibiotic were placed on top. The plates were incubated for 24 h at 37˚C and the diameter of the

inhibition zones was measured after incubation.

Commercially available disks for rifampicin (30 μg) and vancomycin (30 μg) were used as positive

and negative controls based on their ability to inhibit bacterial growth. However, the filter disk for

the novel antibiotic (PPA148 1 μg) as well as for the complex and final formulation, were prepared

in-house. All stock solutions (PPA148 5 mM in DCM, 1:1 complex (0.5 mg/mL) and encapsulated

Fluidosomes (0.5 mg/ml) were sterilized using a UV lamp (Spectroline®, ENF-24C/FE,

Spectronics corporation, Westbury, New York, USA) at 254 nm (ultraviolet germicidal irradiation). A volume equivalent to PPA148 1 μg was placed on the plate, absorbed and evaporated under

reduced pressure. Each disk can absorb a maximum of 10 μL per deposition. Thus, for the

complex- and liposome-impregnated disks the process had to be repeated 5 times because the

stock solution/suspension was not concentrated enough. To validate the preparation method of

the in-house disks, disks containing 30 μg of rifampicin were prepared and the result was

compared to that of the commercially available disk.


71

3.3.10 Dispersion stability of empty fluidosomes

Fluidosomes prepared with the probe sonicator were tested for their stability in terms of

hydrodynamic size. Dynamic light scattering was measured using the experimental parameters described in section 3.3.6. The stability of the dispersion was investigated by measuring their

size over a period of 60 days as a concentrated suspension (1 mg/mL) and diluted in PBS buffer,

pH 7.4 (dilution factor of 2, 4, 8 and 16). The effect of adding HPβCD (0.1, 1, 2.5 and 5%) on the

size of the liposomes (1 mg/mL) was also examined. All samples were kept at 4˚C between the

measurement intervals and were measured in triplicates.

3.3.11 Statistical analysis

All error values were expressed as their mean ± standard deviation (SD). Statistical analysis of

the data was performed using GraphPad Prism 7. The non-parametric Mann-Whitney U-test was

used to detect differences between the empty and loaded liposomes. Statistically significant

differences were assumed when P ≤ 0.05. For the disk diffusion assay, simulations of each data

set were computed to tabulate 100 runs based on a Gaussian absolute scatter and the standard

deviation of each experimental data set. The Brown-Forsythe variability test was conducted to

ensure all sample group data was of acceptable distribution (P > 0.05) before statistical significance between the groups was assessed by one-way analysis of variance (ANOVA) tests

with post-hoc Tukey analysis for multiple comparison. Statistically significant differences were

assumed when P ≤ 0.05 for the experimental analysis and the simulated observations. P-value

style is based on GraphPad Prism 7 and the asterisks symbolize the significance as follows:

0.1244 (ns), 0.0332 (*), 0.0021 (**), 0.0002 (***), <0.0001 (****).

For the Kirby Bauer microbiological assay, Monte Carlo simulation was performed for the data

sets because the number of observations was small, and any outlier could affect the final result.

A number of 100 simulated observations were generated based on the mean and standard

deviation of the experimental data set. These simulated data were used to perform one-way

ANOVA analysis in addition to Tukey’s test as described above, which give an estimation of the

result if we had 100 experimental observations.

3.4 Results

3.4.1 Characterization of empty/control Fluidosomes

Liposomes consisting of DPPC/DMPG in a molar ratio of 18:1 were manufactured using the thin film hydration method. Size reduction was performed by either probe sonication of the lipid

suspension or extrusion of iced/thawed lipid suspension through a polycarbonate membrane of

100 nm pore size. The lipid suspension was applied to a gravity column, washed 8 times and the

eluted aliquots were collected. The column fractions were tested for the presence of liposomes


72

by measuring the derived count rate and the lipid content by the Stewart assay. The aliquot

containing liposomes was further characterized for liposome size and zeta potential.

Dynamic light scattering was initially used to verify the presence of liposomes after the

manufacturing and purification process. The derived count rate (DCR) represents the average

intensity scattered and is directly related with the number of photons detected and, thus, the

concentration and number of particles present in the eluted samples from the PD10 column

suspension. The development of the Fluidosome preparation began in parallel with the synthesis

and characterization of the drug, described in Chapter 2. During this period, small unilamellar vesicles (SUVs) were prepared by probe sonication and to test their stability over time and

dilution. This method produced liposomes (1 mg/mL) with size of 56.30 ± 0.50 nm and PdI of

0.21 ± 0.01 (Figure 3.2A). The size of the 1 mg/mL liposomes was also tested over time (Figure

3.2A). The shift of the intensity distribution towards higher hydrodynamic size reflects an increase

in liposomal size of 27.3 ± 2.5% during a period of 2 months (Figure 3.2A), which suggests some

aggregation of the liposomes (189).

The effect of PBS on the stability of Fluidosomes was examined by diluting the initial lipid

suspension up to 16 times from its original concentration (1 mg/mL). The size of all 4 diluted

liposomal suspensions was measured and compared over time (Figure 3.2B). Between

measurements, the samples were stored at 4 ˚C and the measurements were performed at 25

˚C. The least concentrated suspension (0.0625 mg/ml) displayed a 16.0 ± 3.1% and 55.3 ± 2.6%

increase 4 days and 2 months after fabrication, respectively (Figure 3.2B). The liposome size in

the rest of the samples exceeded the original size by 10%, 11 days after preparation. Their overall

size increase ranged between 19 and 29% 1 month after preparation (Figure 3.2B). The results

on the stability of liposomes are in agreement with studies on DPPC liposomes at the same

concentration (1 mg/mL) (191).


73

Figure 3.2: Empty Fluidosomes (SUVs, 1mg/mL) were manufactured with the thin film hydration method and probe sonication for size reduction. (A) Effect of liposome integrity in terms of size and intensity distribution profile of liposomes at a concentration of 1 mg/mL and temperature of 25 ̊ C, measured by dynamic light scattering, over a period of 2 months. The samples were stored at 4 ˚C and tested between measurement intervals. All samples presented a polydispersity index of approximately 0.2 ± 0.01. (B) Effect of PBS buffer pH 7.4 on the size of a series diluted SUVs (1, 0.5, 0.25, 0.125 and 0.0625 mg/mL) over time.

Since the objective was to produce drug/CD inclusion complexes-in-liposomes, the stability of

the fluidosomes in the presence and absence of cyclodextrin (HPβCD) over time was also investigated. Liposomes, made in the presence of 0.1, 1, 2.5 and 5% w/w HPβCD, produced

particles with a size of 67.70 ± 0.40, 54.10 ± 0.50, 56.50 ± 0.50 and 53.00 ± 0.10 nm, respectively

(Figure 3.3A). It is important to mention that liposomes were not purified, therefore, cyclodextrin

was present both in the bulk and encapsulated in the liposomes. The addition of cyclodextrin did

not affect the size of the original fluidosomes (56.30 ± 0.50 nm). It can be seen from Figure 3.3

that fluidosomes exhibited a higher stability when HPβCD was present. The size increase did not

exceed 5% of the initial size over the period of 2 weeks studied. However, the polydispersity

index was high (approximately 0.30 for all samples), compared to empty fluidosomes (0.21 ±

0.01), and did not change by increasing the amount of cyclodextrin. Despite the narrowing of the

size distribution by adding 5% HPβCD, the polydispersity index remains high (0.26 ± 0.01). It is

observed that the 0% HPβCD in Figure 3.3A presented a broader and not a unimodal distribution

compared to the equivalent 1mg/mL empty fluidosome sample in Figure 3.2A. The probe

sonication as a size reduction method did not present reproducible liposomal size distribution.


74

Figure 3.3: Fluidosome (1 mg/mL SUVs) encapsulated with a series of HPβCD solutions in PBS buffer pH 7.4 (0.1, 1, 2.5 and 5% w/w) manufactured with thin film hydration method and probe sonication for size reduction. (A) Effect of HPβCD on the size and polydispersity of Fluidosomes in terms of the intensity distribution profile of liposomes at 25 ˚C, measured by dynamic light scattering. All samples presented a polydispersity index of approximately 0.3 ± 0.01. (B) Comparison of size changes in the absence and presence of HPβCD.

Despite the fact that SUV DPPC liposomes are inherently unstable compared to LUVs (190,191),

the liposomal size distribution was not reproducible, the polydispersity index was often high and

the frequent release of titanium particles into the lipid suspension from the probe-sonicator led

us to a new approach for size reduction of liposomes with improved size distribution: the extrusion

method, which uses polycarbonate membranes of 100 nm pore size, and produces liposomes

with sizes of 100-130 nm (large uni-lamellar vesicles, LUVs). It is also expected that liposomes around 100 nm in diameter are optimum for drug delivery into bacterial cell membranes, since

they can trap higher volumes, while preserving high surface tension required to fuse with bacterial

membrane (192–195), a size double the one we obtained with probe sonication. The extrusion

of liposomes was performed twice with an incubation time of 12 hours in between. After the first

extrusion, the light scattering data showed a multimodal size distribution. This profile was

transformed into a monomodal distribution as presented in Figure 3.4B and Table C 1 when the

already extruded lipid suspension was extruded again after 12 hours. Liposomes of desired size

(100 nm) were successfully formed and passed through the size exclusion column. The majority

of the particles were eluted in the third fraction with a hydrodynamic size diameter of 148 ± 19

nm and a polydispersity index of 0.33 ± 0.07 (Figure 3.4B). Even though the polydispersity index

remains high, the size distribution is narrower and uniform (Figure 3.4B) when compared to that resulting from probe sonication (Figure 3.2A).


75

Figure 3.4: Method development for the separation of unentrapped material from fluidosomes. Empty liposomes (2.5 mg/mL), prepared with thin film hydration method and extrusion through a 100 nm polycarbonate membrane for size reduction, were passed through a Sephadex PD10G25 size exclusion chromatography column, washed 8 times and the eluted volumes (8 fractions) were collected and analyzed. (A) Elution pattern of empty fluidosomes (2.5 mg/mL). The derived count rate profile and lipid content in each aliquot were measured to investigate the elution of liposomes. (B) Representative intensity distribution of empty fluidosomes in HEPES buffered saline at pH 7.2 (aliquot #3).

The Stewart assay was performed to determine the lipid concentration in each aliquot, which

confirmed that liposomes were eluting in the third aliquot, as also shown from the DCR data

(Figure 3.4A). In the third fraction, 1.39 ± 0.02 mg/mL lipids were measured, which represents

55 ± 2% of the initial lipid suspension concentration (2.5 mg/mL). Overall, there was a 29.4 ±

3.7% loss of lipids due to the separation column because the initial lipid concentration agrees

with the amount of lipids used to make liposomes.

3.4.2 Characterization of drug-CD inclusion complexes in the Fluidosomes

For the fabrication of fluidosomes encapsulated with the drug/CD complex, RAMEB was used because PPA148 showed a higher affinity towards this type of cyclodextrin (Chapter 2). From

this point on, RAMEB was used for complex formation in the experiments. A 1:1 complex of

PPA148/RAMEB was incorporated into the fluidosomes, and the elution profile of the loaded

liposomes (Figure 3.5A) was the same to the one with the empty liposomes (Figure 3.4A, Table

C 1). The majority of fluidosomes eluted in the third fraction were of a size of 129 ± 10 nm, which

was not significantly different from that of the empty liposomes (P = 0.14). A low polydispersity

index (0.14 ± 0.03, compared to 0.33 ± 0.07 for the empty liposomes) was observed, indicating

that the encapsulated drug/cyclodextrin complex improved the monodispersity after the second

extrusion. The guidelines for manufacturing liposomes given by AVANTI Polar Lipids additionally

recommend to keeping the lipid suspension overnight (aging process) prior to downsizing

because it improves the homogeneity of the size distribution (190).


76

Figure 3.5: (A) Elution pattern of PPA148/RAMEB encapsulated fluidosomes (2.5 mg/mL). The derived count rate profile and lipid content in each aliquot were measured to investigate the elution profile of loaded liposomes. (B). Representative intensity distribution of drug/RAMEB encapsulated fluidosomes in HEPES buffered saline at pH 7.2 (aliquot #3).

The third fraction contained liposomes with ζ-potential close to neutrality (+3.4 ± 0.3 mV), and

held 77 ± 22% of the initial lipid concentration (10 mg/mL), with an overall lipid loss of 13 ± 9%,

as assessed by the Stewart assay. Lipid loss due to the extrusion process was not observed (the amount of lipids in the liposomes before separation matches the starting lipid concentration for

all batches tested).

3.4.3 Encapsulation Efficiency and Drug Loading

The eluted aliquots of the loaded fluidosomes were tested for their drug content. Each aliquot

was lyophilized and re-suspended in EtOH/H2O (80/20), which was shown in Chapter 2 to be

the best solubilizing solvent mixture for PPA148 (Figure 2.13). Their UV/Vis spectrum was recorded. The separation of free drug from the formulation was carried out using size exclusion

chromatography as mentioned in section 3.3.5. The drug quantification profile of the eluting

fraction is shown in Figure 3.6A. It was performed using the area under the curve (AUC) of the

spectra ranging from 250-400 nm. This approach was chosen because the spectra of the

extracted PPA148 did not present reproducible spectra in terms of peak shape and position. In

Figure 3.6B, the UV spectrum of pure PPA148 in ethanol/water is compared to that of extracted

PPA148 from aliquot 3, lyophilized and resuspended in ethanol/water. Pure PPA148 presents two peaks at 295 and 275 nm (Figure 3.6B). PPA148 from aliquot 3 produced either sharper

peaks and a shift in λmax, or one obscure broad peak instead of two and a high baseline even

after subtracting the solvent spectrum (Figure 3.6B). The hyperchromic shifted baseline at the

red end of the spectrum probably reflects the presence of scattered light which may be a result

of incomplete drug extraction leading to the presence of aggregated cyclodextrin and/or lipid

particles. Any remaining lipid and CD in the test samples introduced a variety of artefacts and it


77

was difficult to quantify the drug based on λmax. In total, four batches were prepared and tested

and three had the version 2 profile shown in Figure 3.6B.

The alternative method of the area under the absorbance-wavelength curve (AUC) was

considered, as described by Zhang et al. (81). This approach was investigated by building a

calibration curve and it was verified that the AUC is proportional to drug concentration (inset of

Figure 3.6A) which provided a robust method for the drug quantification.

Based on the separation of the fluidosomes from the unentrapped drug/CD, liposomes (empty

and containing the drug/CD complex) eluted in the third fraction of the column. The highest drug

content (280 ± 120 μg/mL) was found in the same fraction with encapsulation efficiency (EE) of

67 ± 11% and drug loading (DL) of 5 ± 1%, which were calculated based on Equating 3.2 and

3.3. The EE and DL values of the DPPC/DMPG liposomes employed in this work were consistent

with those previously reported using different drugs but similar preparation methods (196,197).

The total drug concentration (590 ± 113 μg/mL) obtained from the total eluted volume from the

Sephadex column is in agreement – within experimental error - with the initial drug concentration (500 μg/mL) which was used to hydrate the lipid film.

Figure 3.6: (A) Drug quantification profile of the eluted fraction from the PD10G25 size exclusion chromatography column with the inset being the calibration curve (y=0.7937+5.717,R^2=0.9966) is presented as a function of the area under the curve (AUC). (B) The UV spectra of pure and extracted drug in ethanol/water (80:20). Version 1 and 2 and the two UV spectra of extracted PPA148 from aliquot 3, which present changes in drug peak shape.

3.4.4 Kirby Bauer assay

The efficacy of the encapsulated drug/RAMEB complex in liposomes was tested by using the

disk diffusion assay. Commercially available disks of rifampicin and vancomycin were used as

positive and negative controls against E. coli DH5a. Rifampicin causes inhibition of bacterial

growth, while vancomycin does not affect the growth, resulting in the absence of an inhibition

zone. Disks impregnated with PPA148 (1 μg/mL), rifampicin (1 and 30 μg/mL), RAMEB,


78

PPA148/RAMEB complex, empty fluidosomes, and encapsulated fluidosomes with

PPA148/RAMEB were prepared in-house. In-house rifampicin disks (30 μg/mL) were compared

with the commercially available disks of the same concentration to validate the preparation method (data not shown). Figure 3.7 presents the inhibition zone caused by the different systems

studied. Rifampicin (30 μg) caused inhibition of bacterial growth, as expected, and was used as

a positive control, while vancomycin (negative control) did not affect it.

Figure 3.7: Kirby-Bauer assay for measuring the growth inhibition of Escherichia coli DH5α. The inhibition zone diameter was measured for PPA148 as a pure substance and as a formulated drug (in a 5% RAMEB complex and incorporated into 0.1 mg fluidosomes as a complex with 5% RAMEB) after 24 hour incubation at 37 °C. Rifampicin and vancomycin were used as positive and negative control samples. The asterisks denote the level of significance going from lower to higher (as follows) and four asterisks indicate the highest level: 0.1244 (ns), 0.0332 (*), 0.0021 (**), 0.0002 (***), <0.0001 (****). All samples were tested in triplicate apart from the final formulation (PPA148-in-RAMEB-in-fluidosomes) which was tested 5 times.

One-way ANOVA test showed that, overall, the groups were statistically different (P=0.0007) at

90% confidence interval. The multiple comparison showed a clear picture within each group.

There was a statistically significant difference between the free drug, RAMEB/PPA148 complex

(P=0.0041) and PPA148-in-RAMEB-in-liposomes(P=0.0006). However, the difference between

the means of the drug/CD complex alone and those encapsulated within liposomes was not

statistically significant (P=0.5483,) (Figure 3.8A). Tukey’s test compares the means of all groups

to the mean of every other group and is considered the best available method when confidence intervals are desired or if sample sizes are unequal (Figure 3.8B). The interval of PPA148-in-CD

and PPA148-in-CD-in-liposome contains the number zero which indicates that the means of

those groups is likely to be the same (Figure 3.8A). The residuals denote the difference of each

observation of the mean (Figure 3.8B). They are evenly distributed, indicating a good and reliable

set of data.


79

Figure 3.8: Tukey’s test of experimental data (n=3,5) with confidence interval of 95%. (A) Zero is included in the interval of PPA148-in-CD and PPA148-in-CD-in-liposome which indicates that their mean is not statistically significant. Residuals distribution of experimental observations for each group (B).

However, the number of observations is too small (n = 3 or 5) and any outliers may affect the

result. Therefore, Monte Carlo simulations were conducted. A large number of observations (100) was generated for each group based on the experimental mean and standard deviation. The

data sets generated were then used to perform one-way ANOVA analysis and Tukey’s test. The

simulation resulted in a statistical difference among all sets with a confidence interval of 95%

(Figure 3.9A). The P-values for drug vs PPA148-in-RAMEB, PPA148 vs PPA148-in-RAMEB-in-

fluidosomes and PPA148-in-RAMEB vs PPA148-in-RAMEB-in-fluidosomes were <0.0001,

<0.0001 and 0.0002 respectively. This model predicts that more replicates should give a

statistical difference between the intermediate (complex) and final (drug-in-CD-in-liposome) formulation, too.

Figure 3.9: Monte-Carlo simulation was carried out based on the standard deviation of the experimental data. 100 observations were generated by the simulation and one-way ANOVA and Tukey’s test was carried out with 95% confidence interval (A). All sets of groups have statistical difference. Residual distributions of simulated data (B).


80

In summary, PPA148 inhibited bacterial growth, as expected from the MIC results (107). Both

formulations developed, either simply as an inclusion complex with RAMEB or the same complex

in liposomes (DPPC/DMPG, 18:1), which increased the efficacy of this novel agent.

3.5 Discussion

In this chapter, PPA148 was formulated with cyclodextrins and a liposome mixture

(DPPC/DMPG) to enhance its efficacy against Gram negative bacteria. The rationale behind this

formulation is that cyclodextrins (RAMEB and HPβCD) enhance drug aqueous solubility, as

shown in Chapter 2, and potentially enhance transport through the IM, while the liposome is the

drug carrier which will provide an improvement in drug transport through the OM, thus increasing

drug efficacy. Fluidosomes were successfully manufactured using the thin film hydration method,

followed by either probe sonication or extrusion for size distribution. The probe sonication

produced small unilamellar vesicles, because their size is less than 100 nm (56.30 ± 0.50 nm)

and presented low polydispersity index (0.21 ± 0.01) in PBS. A stability study of the SUVs, in

terms of size, showed that the hydrodynamic diameter of the liposomes underwent an increase in size over a two-month time period, with the highest increase being 50% for the most diluted

sample (0.0625 mg/mL) (Figure 3.2). Due to their small size, i.e. high degree of lipid curvature,

DPPC small unilamellar vesicles are inherently unstable and are expected to spontaneously fuse

to form larger vesicles when stored below their phase transition temperature because they are

in their gel-like state (190), which would be the case for the fluidosomes because they were

stored at 4°C in between the measurement intervals.

Apart from the curvature, liposomal stability is dependent on the presence of ions seeing that

they affect the hydration state of lipids and, therefore, their packing. There are two hydration

centers in phospholipids, namely, the phosphate and the carbonyl groups, which determine the

intermolecular lipid interactions (129). Ions and water organization near the different groups

esterified in the phosphate group (choline, ethanolamine or glycerol) affect the hydrophobic and

repulsive forces between the lipid head-groups at the interface (198,199). Sodium (Na+) and

potassium (K+) bind to the carbonyl groups of POPC, which become slightly more accessible to hydration (200) with the first having a stronger binding compared to the latter, due to their smaller

size (201). Although Na+ and K+ (0.15 M) bind to the carbonyl group, they cause almost no effect

on the POPC packing (202). In this study, small unilamellar DPPC/DMPG (18:1) liposomes were

formed in PBS containing Na+ at concentration of 0.2 M, which possibly did not provide size

stability over time (Figure 3.2B).

When a series of HPβCD solutions (1-5% w/w) were used to hydrate the lipid film, the

hydrodynamic size diameter did not change and the stability of the liposomes was higher, as no

significant change in size occurred over a period of two weeks (Figure 3.3). Native and synthetic

cyclodextrins have been reported to interact with lipids and have indeed been used in

pharmaceutical and food products to either stabilize or solubilize bioactive lipids (203). Generally,


81

cyclodextrins interact with phospholipids mainly though the inclusion of fatty acid chains into their

cavity since the polar headgroup interaction with the CD cavity is limited and weak (203). Native

CDs have been reported to form complexes of low solubility with phospholipids while their derivatives present a higher degree of interaction (203). RAMEB (45 mM) has been shown to

solubilize phospholipids (181), while non-methylated CD derivatives and in particular HPβCD do

not increase the permeability of DPPC liposomes for concentrations up to 125 mM (17% w/w)

(181,203–205). In this study, we found, from the evolution on size (Figure 3.3), that HPβCD (up

to 5%) stabilized the fluidosomes rather than solubilizing the lipids as also found by other groups

(181). The interaction of both RAMEB and HPβCD with lipid monolayers (DPPC/DPPG and

POPC/POPG in molar ratio of 3:1) have been studied by surface pressure measurements in a

Langmuir trough and are described in Chapter 4. Our results (Chapter 4) show a weak interaction

between HPβCD and the DPPC/DPPG monolayer, while RAMEB displays a stronger interaction (Figure 4.9).

The instability of SUVs, due to their small size, high curvature and choice of buffer, led us to

explore an alternative approach to manufacturing large unilamellar vesicles (LUVs) using

extrusion, to achieve sizes above 100 nm. LUVs display higher stability and retain the large

encapsulation efficiency of SUV (115,190). LUVs of a hydrodynamic diameter of 148 ± 19 nm

and a polydispersity index of 0.33 ± 0.07 (Figure 3.4A) were prepared in HEPES buffered saline

containing 20 mM CaCl2. Ca2+ were added into the saline (145 mM) because they tend to stabilize

unilamellar lipid PC vesicles at a concentration range of 1-60 mM by forming Ca2+ bridges between lipids (206). Ca2+ ions at low concentration (1-5 mM) adsorbs deeply into POPC and

POPS bilayer (202) and they bind more strongly to the carbonyl, phosphate and carboxylate

groups of both PC and PG, compared with Na+ ions, causing partial dehydration, conformational

change and immobilization of the phosphodiester groups (200,207). The influence of 1 mM CaCl2

is comparable to the one obtained with 150 mM NaCl, which shows the different affinity of each

ion to the lipid headgroup (202). The increased size of LUVs, along with the presence of Ca2+,

creates a more stable environment for fluidosomes. The total ionic strength of HEPES buffered saline containing CaCl2 was 211 mM which is of the same magnitude as PBS used in the SUV

preparation. According to Lapinski et al. (208), if the ionic strength of the two buffers is the same,

experimental data on bilayer lipid membranes formed by sonication (SUVs) can, in fact, be

compared directly to data on bilayer systems formed by extrusion (LUVs). Based on that finding,

the presence of HPβCD could possibly stabilize LUVs as found in the case of SUVs (Figure

3.3B).

The LUVs were encapsulated with the PPA148/RAMEB complex and it was observed that the

size of the encapsulated liposomes (129 ± 10 nm) was not statistically different from that of the

empty liposomes (148 ± 19 nm). The stability of the loaded liposomes depends on the competitive

binding of either the lipids or the drug (PPA148) to the CD cavity. This may happen because

inclusion complexes in solution are in equilibrium with the free molecules (209) and have been


82

found to extract phospholipids and cholesterol from membranes by forming complexes (181). In

Chapter 2 (Figure 2.17A), the affinity of PPA148 to RAMEB was calculated by measuring the

binding constant using fluorescence spectroscopy. The binding constant of DPPC/RAMEB complexes has not been measured and no reference was found in the literature. Instead, the

binding constant for a 1:1 cholesterol/DIMEB complex is characterized by a binding constant of

102 M-1 (210), which is similar to the binding constant of drug/RAMEB (102 ± 26 M-1) found in

Chapter 2. As cholesterol has higher affinity towards cyclodextrin methyl-derivatives than DPPC,

it can be speculated that DPPC undergoes weaker binding with RAMEB than PPA148. The

presence of drug/CD complex improved the polydispersity index of the lipid suspension,

indicating the formation of a more monodisperse system. RAMEB was used at low concentration (0.7 mM) as it is known to solubilize phospholipids above 10-15 mM membrane (181,182). Below

this cut-off concentration, RAMEB retains the ability to facilitate lipid exchange without disturbing

the membrane (181,182).

The next step was to separate the loaded liposomes from the unentrapped material using size

exclusion chromatography and characterize the system in terms of encapsulation efficiency and loading capacity. The separation method was first validated using unloaded empty fluidosomes

to develop a robust eluting pattern of the fluidosomes. The lipids were eluted from the Sephadex

column by 8 successive additions of HEPES buffered saline pH 7.2 after equilibrating the column

with 15 mL of buffer. After the separation, an overall loss of lipids of 29.4 ± 3.7% and 13 ± 9%

was observed for the empty and encapsulated liposomes, respectively. The third aliquot of the

eluted empty fluidosomes contained 55 ± 2 % lipids of the initial amount used, as assessed by

the Stewart assay (Figure 3.5). The rest (29% lipid loss) was probably retained in the column

because lipid loss was not observed during the extrusion stage. When using the liposomes

prepared in the presence of the drug/RAMEB complex, the loss of lipids after the separation

process was lower, at only 13 ± 9% (Figure 3.5). The loss of lipids may be a result of poor column

condition before elution. It has been reported elsewhere (211) that pre-saturation of the column

with empty liposomes helps to avoid material loss during size exclusion chromatography. The

binding and release of liposomes on the dextran stationary phase of a Sephadex column is a

dynamic equilibrium process between the free and entrapped liposomes. However, pre-

saturation of the column can be problematic as it may contaminate the test sample. Further

investigation needs to be done in order to clarify the cause of lipid loss during the size exclusion chromatography.

The high encapsulation efficiency of drug in the liposomes, assessed by the area under the

absorbance-wavelength curve (67 ± 11%) showed that the drug/RAMEB complex was

successfully incorporated into the liposomes. Drug quantification was performed after extracting

PPA148 from the carrier and cyclodextrin, by washing with DCM. The extracted PPA148

displayed a UV spectra with λmax shifted compared to the pure drug in ethanol/water and change

in the shape of the peak (Figure 3.6B). Based on these data and the findings in Chapter 2 of a


83

PPA148/RAMEB complex, it was assumed that the extracted PPA148 from the formulation was

not pure or had undergone structural conformation changes, leading to an altered UV profile

(Figure 3.6B). It is possible that the extraction was incomplete because of incomplete dissociation of the PPA148/RAMEB complexes, the presence of cyclodextrin dimers or even RAMEB/lipid

complexes that had not been removed from the mixture. Yamamoto and coworkers investigated

the dissociation of cholesterol/βCD complexes and cholesterol recovery was found to be 44%

(212). They found that dissociated and undissociated cholesterol was present in the mixture,

along with βCD dimers. The encapsulated fluidosomes with PPA148/RAMEB complexes is a

very complex system and during the drug extraction several types of aggregates may have been

formed, affecting the final drug absorbance. RAMEB may associate with lipid headgroups while being complexed with PPA148 as found with mono–(N-n-alkyl,N,N-dimethylamino)-β-

cyclodextrin, which can associate with the bilayer of DPPC:cholesterol (7:3) liposomes, while

having the adamantoyl moiety of the adamantoylglucose molecule included into the cavity (213).

Since the size of the encapsulated fluidosomes did not change and the encapsulation efficiency

is high, possibly PPA148 may be inserted into the CD cavity from one side and RAMEB can be

attached from the other side with the headgroups of the inner leaflet of the liposomes (Figure

3.10). Although there is no evidence concerning such an arrangement, it may explain the poor

drug extraction because of the complexity and difficulty in separating the individual molecules.

The efficacy of PPA148-in-RAMEB-in-liposomes was assessed against live bacteria (E. coli,

DH5α) (Figure 3.7). The disk diffusion assay revealed a statistically significant increase in the

inhibition zone of the drug-CD complex on its own and when it is incorporated into liposomes.

Due to the small number of observations (n=3-5) the difference between the PPA148/RAMEB

complex and the encapsulated liposomes was not significant (Figure 3.8). Based on the fact that cyclodextrin enhance the antimicrobial activity of drugs (171) and the lack of significance

between the PPA148/RAMEB complex and the encapsulated fluidosomes, Monte Carlo

simulations were conducted in order to investigate this result further. A higher number of

simulated observations (100) were generated based on the standard deviation of the

experimental data sets and tested using one-way ANOVA. The results showed a statistical

difference between the drug/RAMEB and the drug/RAMEB incorporated into liposomes (Figure

3.9). The increased efficacy of antibiotics by encapsulation into liposomes was first reported by

Beaulac and co-workers in 1998 and later by Sachetelli et al. in 2000 (102,179). They showed that DPPC/DMPG (18:1) liposomes encapsulated with tobramycin showed a decrease in the

bacterial counts in a sub-MIC concentration (102,179). Other reports regarding increased

efficacy of antibiotics with fluidosomes and other types of liposomes have been published up to

the present time (98). Interestingly, there have been publications regarding the increased efficacy

of drugs in the form of cyclodextrin complex or encapsulated into liposomes (Table 1.3, Table

1.4) (90,214), but there is only one report about a drug-in-CD-in-liposomes formulation in the

food industry (113), with nerolidol, a sesquiterpene with antibacterial activity and under investigation for its potential as a natural preservative in the food industry. In the next Chapter,


84

we use lipid monolayers representing the inner and outer bacterial membrane and a model Gram

negative bacterial outer membrane to elucidate the mechanisms responsible for the increased

activity of this novel antimicrobial agent by using interfacial techniques (Langmuir trough monolayer interaction studies and neutron reflectivity).

Figure 3.10: Schematic representation of the possible organization of PPA148/RAMEB complex incorporated into Fluidosomes. The electrophilic center of PPA148 is presented inside the CD cavity while the headgroup of DPPC is associated with the exterior hydrogens of the narrow side of RAMEB molecules.

3.6 Conclusion

Fluidosomes encapsulated with PPA148/RAMEB complex were successfully manufactured with

a size of 129 ± 10 nm, polydispersity index of 0.14 ± 0.03 and high encapsulation efficiency (67

± 11%). The separation method of unentrapped material from loaded liposomes was developed

using empty liposomes. The final formulation was tested against E. coli DH5α and presented a

significant increase in drug efficacy. In the next chapter, PPA148 in its pure form and

cyclodextrins (HPβCD and RAMEB) are assessed over a series of lipid monolayers. The effect

of fluidosomes on the outer membrane is tested on a model Gram negative bacterial membrane,


85

which was developed at the ISIS neutron source (Didcot, Oxfordshire) and neutron reflectivity is

used to assess the interaction.

A biophysical investigation into the uptake mechanism of PPA148 and its delivery system

86

Chapter 4 A biophysical

investigation into the uptake

mechanism of PPA148 and its

delivery system


87

4.1 Introduction

The uptake of drugs and their delivery systems across the bacterial cell envelope is a complex

process, which is difficult to elucidate because of its dynamic nature. The envelope of Gram

negative bacteria possesses two membranes, acting as discrete permeability barriers which

drugs and nutrients need to cross so as to enter the cytoplasm. The outer membrane (OM) of

Gram negatives consists of an asymmetric lipopolysaccharide (LPS) and phospholipid bilayer spanned by porin proteins which play an important role in nutrient and drug uptake (Figure 1.2).

The β-barrel porins are transmembrane proteins, which regulate the uptake of nutrients and

hydrophilic compounds. The OM also provides protection against environmental stressors and

inhibits the uptake of some antibiotics. The second permeability barrier, the inner membrane

(IM), consists mostly of intrinsic proteins and phospholipids. The uptake of compounds by the

bacterial cell is divided into passive and active processes. The former is characterized by either

drug penetration through the lipids or transport through general or specific transporter proteins. The latter requires energy input for the transport of compounds through the proteins. The

transmembrane proteins (porins), located in the OM, can act as both passive and active

transporters for the influx of beneficial compounds and efflux of unwanted molecules, such as

antibiotics. The nature of LPS and the presence of porins challenge the antibiotic uptake by Gram

negative bacteria and there is a real need to find means of overcoming these drug permeability

obstacles.

In Chapter 3, a drug-in cyclodextrin-in liposome formulation was evaluated, which showed an

enhancement of PPA148’s solubility and antimicrobial efficiency. The rationale behind this

formulation is based on two factors: the results from the microbiological assay (Table B 1; Table

B 2) of pure PPA148; and PPA148’s physicochemical properties, in particular water solubility

and lipophilicity, as calculated by ChemDraw software (Table 1.5). The results from the

microbiological assays (minimum inhibitory concentration or MIC) in the presence and absence

of the efflux pump inhibitor phenylalanine-arginine-β-naphthylamide (PAβN) (107,108) revealed difficulties in the transport through the cell envelope and a possible synergistic effect between

PPA148 and PAβN either by inhibiting the efflux pumps or by PAβN's permeabilizing effect on

both OM and IM (69). The MIC, in the absence of the pump inhibitor, was relatively high (2 μg/mL

against A. baumannii, 0.25-32 μg/mL in K. pseumoniae and 16-128 μg/mL in P. aeruginosa) but

decreased dramatically in all bacterial strains tested in its presence (107). In this work, a

liposomal carrier was used to overcome PPA148’s uptake challenges. The microbiological assay

described in Chapter 3 showed an enhanced inhibition of E. coli DH5α bacteria when the

drug/RAMEB complex was incorporated into liposomes (Figure 3.7; Figure 3.8; Figure 3.9). Fluidosomes were selected to enhance the transport of PPA148 through the OM because they

have been found to fuse through bacterial membranes (179). It was, therefore, hypothesized that

the fluidosomes would fuse into the OM while releasing their core content into the periplasmic

space. Based on the lipophilicity of PPA148, it was also hypothesized that the novel drug might


88

partition into the IM in order to reach its site of action inside the cell. The low water solubility of

PPA148 may lead to drug precipitation when entering the periplasm, leading to ineffective

antimicrobial action. For this reason, RAMEB was successfully used as a solubilizing agent for PPA148 (Chapter 2), which is barely soluble in aqueous solvents as a pure substance. It has

also been reported that β-cyclodextrins and their derivatives, such as RAMEB and HPβCD, not

only enhance water solubility of poorly soluble drugs but also increase their internalization into

cells and improve their biological activity against Gram negative bacteria (90,91,94). RAMEB has

been found to cause cell lysis and inhibit the growth of the Gram positive Bacillus strains (185).

While the mechanism of bacterial growth inhibition by RAMEB has not been investigated, native

βCD has been found to adhere to the peptidoglycan surface of Gram positive bacteria via hydrogen bonds (215) and to be transported with facilitated diffusion via specific channels

(CymA) in the outer membrane of the Gram negative bacterium, Klebsiella oxytoca (31). In Gram

negative bacteria, increased electrostatic interactions have been observed between the inclusion

complex of chlorhexidine/native-βCD and the LPS of Aggregatibacter actinomycetemcomitans,

whose structure is similar to that of E. coli (216,217). Therefore, we hypothesized that

cyclodextrin may also help the passive transport of PPA148 through the IM via the formation of

phospholipid-cyclodextrin complex by releasing pure PPA148 inside the cytoplasm where the

site of action (DNA minor groove) is located.

Based on the results from Chapters 2 and 3, this chapter investigates the mechanism of uptake

of this novel antimicrobial agent and assesses the effectiveness of a drug-in-cyclodextrin-in-

liposome formulation to increase OM and IM permeability. The investigation of the drug-

membrane interactions and possible drug penetration was carried out by using interfacial

techniques (Langmuir trough and neutron reflectivity) with both monolayer and bilayer model membranes. Passive diffusion can occur either via diffusion across the lipid bilayers for

antibiotics with a high degree of lipophilicity and some degree of polarity or via the OM proteins,

which can either be specific transporters or general water channels, for small hydrophilic

compounds (218). Based on the possible fusion mechanism of fluidosomes, the cyclodextrin-

phospholipid complex formation and the high lipophilicity of PPA148, simplified artificial

membrane systems were used to mimic the lipidic component of the relevant biological

membranes in order to examine the passive diffusion across lipid layers.

In the monolayer studies, lipids representing the outer leaflet of OM and IM of the Gram negative

bacterial envelope were used to model the membranes. There are two types of LPS: rough and

smooth, which differ by the presence or absence of O-antigen. Mutations into the bacterial

genome produces LPS without or with truncated O-antigen which are called rough or semi-rough

LPS chemotypes. Within the rough chemotypes, further modifications and mutations lead to the

expression of LPS with shorter oligosaccharide, divided into Ra, Rb, Rc, Rd and Re subgroups, depending on the length of their core region (Figure 1.10). Ra LPS has the longest and Re LPS

has the shortest core. In this study, the focus is on a Ra-EH100 and Rc-J5 LPS extracted from

E. coli and Lipid A extracted from the Re-R595 S. minessota. In addition, Rc LPS was used to


89

form stable planar monolayers because of its short headgroup. This type of LPS and Lipid A

extracted from Re LPS were also used to test the influence of the OM steric barrier. The bilayer

studies were conducted on a model asymmetric Gram negative bacterial membrane consisting of Ra LPS as the outer leaflet. Ra LPS is a substitute for LPS molecules containing O-antigen

because smooth LPS is less hydrophobic than truncated LPS and, thus, Ra LPS does not form

water soluble aggregates. It has been found that truncated rough LPS can be deposited as

insoluble monolayers (219,220) while smooth LPS act as surfactants by forming micelles in

solution (221). The more truncated rough mutant LPS was used to form more stable monolayers

and compare the drug-membrane interaction in the presence (Rc LPS) and absence (Re Lipid

A) of steric hindrance.

The phospholipids in the OM are zwitterionic and negatively charged and their distribution within

the individual OM and IM is inhomogeneous (222). The predominant phospholipids are

phosphatidylethanolamine (PE), phosphatidylglycerol (PG) and cardiolipin (CL), while some

species, such as P. aeruginosa, can also produce phosphatidylcholine (PC) (223). It has been

shown that PE and PG are mostly located in the inner and outer leaflet of the plasma membrane (IM) respectively, while CL is evenly distributed in both leaflets of IM (224). E. coli polar lipid

extract has been extensively studied and its phospholipid composition has been found to be 67%

PE, 23.2% PG and 8.9% CL (127). The anionic phospholipids (PG and CL) of the IM represent

18% of the total IM lipids, while they constitute only 9% of the OM lipids. The remaining

phospholipid lipid fraction (>98%) in both the inner leaflet of the OM and the whole IM is PE

(222).

The structural complexity of LPS and the small size of bacteria make it difficult to obtain detailed

molecular information on the interactions between drug molecules and the OM and IM (225).

However, biophysical studies of isolated phospholipid and LPS monolayers and bilayers provide

an insight into these interactions.

In this work, monolayers composed of the phospholipids or rough mutant LPS were deposited at the air/liquid interface and examined in the presence of PPA148 and cyclodextrin using a

Langmuir trough. These experiments provided useful insights into the molecular interactions of

the drug on the membranes, either at the level of the polar region when they are in contact with

the outer leaflet of the OM or IM of the cell envelope, or at the level of the hydrophobic domain

of the lipid bilayer in which they can be partitioned and inserted.

To complement these monolayer studies, a model Gram negative bacterial asymmetric

membrane consisting of Ra LPS and DPPC as the outer and inner leaflet of the OM was used to

examine the interaction between the liposomal carrier and the OM. This model membrane has

been extensively studied and validated by a group of scientists working in collaboration with the

Science and Technology Facilities Council (STFS), Rutherford Appleton laboratory in Chilton,

UK (225–227). Neutron reflectivity (NR) was used to characterize the structural changes in this


90

model membrane when put in contact with the formulation. In particular, NR can give information

on the relative location of fluidosomes within and at the interface of a supported asymmetric

model OM, thus giving clues on the mode of interaction, namely either fusion into the OM or a possible partitioning into the membrane’s hydrophobic core and diffusion across the membrane.

4.2 Materials

Lipids: The phospholipids 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC, 16:0 PC), 1,2-

dimyristoyl-sn-glycero-3-phosphoglycerol, sodium salt (DMPG, 14:0 PG), E. coli polar lipid

extract (E. coli B), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC, 16:0-18:1 PC) and

1-palmitoyl-2-oleoyl-sn-glycero-3- phospho-(1'-rac-glycerol) (sodium salt) (POPG, 16:0-18:1 PC)

powders were manufactured by Avanti Polar Lipids (Alabama, USA) and supplied by Stratech

Scientific (Newmarket, UK). Rc LPS from E. coli J5 (purity: protein 1.4%, nucleic acid 0.340%, phosphate 7.3%, Kdo 5.9%), Ra LPS from E. coli EH100 and Re Lipid A from S. minnessota

R595 containing containing ≤0.3% proteins was obtained from Sigma-Aldrich, UK. and were used

without further purification.

Cyclodextrins: Hydroxy-propyl-beta-cyclodextrin (HPβCD) and randomly methylated

cyclodextrin (RAMEB) were purchased from Sigma-Aldrich, UK.

Chemicals and material used in the microbiological assay: Blank, rifampicin (30 ug) and

vancomycin (30 ug) susceptibility disks, agar powder and Muller Hinton broth for microbiology

were purchased from Oxoid (UK).

Solvents: Dichloromethane (DCM) and chloroform (CHCl3) were purchased from Sigma-

Aldrich, UK. LC/MS grade water was supplied by Merck, UK. The ultrapure water at 18.2 MΩ cm

was produced by a Purelab Ultra machine from ELGA Process Water (Marlow, UK).

Salts used for buffer preparation: Sodium chloride (NaCl), magnesium chloride (MgCl2),

monosodium phosphate (NaH2PO4), sodium hydroxide (NaOH) and HEPES (>=99.5%) were purchased from Sigma-Aldrich, UK.

Devices and tubes: Disposable semi-micro dynamic light scattering cuvettes (volume capacity

of 1.5 mL) purchased from VWR International, UK. Silicon crystal (50 * 80 * 20 mm) polished with

silicon oxide and a customized cell made of PTFE was kindly provided by the ISIS facility (Chilton,

UK).

4.3 Methods

4.3.1 Surface Pressure-Area isotherms at the air-liquid interface

Pressure-area measurements were carried out at a constant temperature of 23°C on a Nima

602A Langmuir trough (Nima Technologies Ltd., Coventry, UK) equipped with a Nima PS4


91

surface pressure microbalance (0-240 mN/m range, 0.1 mN/m resolution) controlled by a PC via

a Nima IU4 computer interface unit. The microbalance was calibrated with a 10 mg standard

weight and the area enclosed by the barriers was also calibrated. Prior to use, the PTFE surface of the trough was thoroughly cleaned with ethanol and chloroform in order to remove any

impurities. A clean Wilhelmy plate (chromatographic paper roll, Whatman Grade 1, 10 mm width,

GE Healthcare life sciences UK, Little Chalfont, UK) was attached to the microbalance,

suspended and partially submerged in the subphase (0.9% w/v NaCl or 1 mM MgCl2). The

subphase surface was examined for dust and contaminants prior to commencing the experiment

by repeated compression of the barriers and suction of the surface using a vacuum pump. The

subphase was considered clean when, upon compression, the change in surface pressure did not exceed 0.2 mN/m.

Solutions of DPPC/DPPG and POPC/POPG mixtures (in a molar ratio of 3/1), Lipid A from S.

minnessota (R595 Lipid A) and LPS J5 from E. coli (Rc J5 LPS) were prepared in pure chloroform

at a total lipid or LPS concentration of 2 mg/mL. To form air/liquid interface monolayers, 30-70

µL of lipid solution were deposited dropwise on the subphase surface using a Hamilton syringe (Hamilton Co. Europe, Bonaduz, Switzerland), with the barriers open at their maximum. Before

compressing the barriers at a constant rate of 35 cm2/min, 10 min were allowed for the solvent

to evaporate. During the compression, changes in the surface pressure were recorded until the

monolayer reached its collapse point. Each sample was run in triplicate. The molecular area for

each monolayer was determined at the pressure of 30 mN/m and compared for their changes in

molecular size. In addition, the compressibility modulus (Es or Cs-1), which is the reciprocal of the

area compression modulus of the monolayer (Cs) (228) , was calculated using the following

formula:

𝐸t =Vah= −𝐴 × ∆�

�� Equation 4.1

, where A is the area per molecule and ΔΠ/ΔA is the slope of the isotherm at a defined surface

pressure (229). The compressibility modulus (Cs-1) of phospholipid and glycolipid monolayers

was calculated at 30 mN/m, which is considered as the approximate lateral pressure of a lipid bilayer (130). The compressibility modulus is an indication of the packing elasticity of the lipid

monolayer and the equilibrium among the lipids. In the liquid expanded film, the compressibility

modulus ranges between 12.5 and 50 mN/m and in for the liquid intermediate takes values bellow

100 mN/m (228). The liquid condensed phase is characterised by a compressibility modulus of

100-250 mN/m while the condensed state, which is described as closely packed lipids in the

equilibrium state, is denoted by values above 250 mN/m (230).

4.3.2 Drug - monolayer interaction

A variation of the adsorption isotherm method (Chapter 2, section 2.3.6) was adopted to

investigate the interaction of drugs and the lipid monolayer. The set-up of the reduced area


92

custom-made Teflon® petri dish was as described in the adsorption studies in Chapter 2 (section

2.3.6). On the clean subphase surface (0.9% NaCl or 1 mM MgCl2) a lipid solution (total lipid

mixture) was spread dropwise until the surface pressure reached 30-35 mN/m, while stirring the subphase with a 4.5×15 mm magnetic stirrer bar. A period of time (40-60 min) was allowed for

the solvent to evaporate and for the monolayer to reach equilibrium (stabilization of the surface

pressure). Thereafter, the drug solution in DMSO was injected below the surface using a small

disposable plastic syringe with a hypodermic needle (BD biosciences UK, Oxford, UK). The

following drug concentrations were used in the trough: 2, 20, 40 and 60 μg/mL when it was tested

with DPPC:DPPG (3:1) monolayer; and 20 μg/mL when it was tested with Rc J5 LPS and Re

595 Lipid A. HPβCD and RAMEB were tested against DPPC/DPPG (3/1), DOPC/POPG (3/1) and Re 595 Lipid A monolayers while fluidosomes were tested at 0.1 mg/mL against J5 LPS

only. The changes in surface pressure were recorded at constant slow stirring of the subphase,

with the magnet speed set at its minimum to avoid a variation of pressure greater than 0.2 mN/m.

The difference in surface pressure change was plotted against time and the binding isotherms

produced were fitted with a sigmoidal model, specifically a Hill plot with 3 parameters as it is displayed in GraphPad, Prism to obtain the kinetic parameters as follows:

𝑦 = ∆�cg�m�

L�z%� pm�

Equation 4.2

,where ΔΠmax is the maximum difference in surface pressure, h is the hill slope, and t50% is the

time needed to achieve half ΔΠmax.

4.3.3 Drug - lipid bilayer interaction using fluorescence spectroscopy

4.3.3.1 Preparation of E. coli B liposomes encapsulating 5(6)-carboxyfluorescein

Liposomes were prepared by the thin film hydration method. A sample of E. coli B lipid extract

(10 mg) was dissolved in chloroform. The solution was placed in a vacuum desiccator overnight

to let the organic solvent evaporate, resulting in the formation of a dry lipid film. The dried film was then hydrated with 2.5 mL of 40 mM 5(6)-caboxyfluorescein (CF) solution (pH = 7.4) at 30

°C. The mixture was vortexed (Cyclone-Vortex mixer CM-1, Nickel-Electro Ltd) for 60 sec at 2800

rpm, sonicated (Soniprep 150 Plus ultrasonic disintegrator 120V 60Hz, London, UK) for 5 min

with an exponential microprobe and amplitude 10 microns and left at room temperature for one

hour prior to further use.

The separation of the liposomes from the unentrapped dye was carried out using size exclusion

chromatography. A Sephadex™ G-25 PD-10 column (GE Healthcare, UK) (packed bed

dimensions of 1.45 × 5 cm and particle size range 85-260 μm) was equilibrated with 15 mL (3

washes of 5 mL) of 70 mM sodium chloride. The liposome mixture (1 mL) was then eluted through

the column and washed with 1 mL of sodium chloride 5 times. The eluted fractions were collected

separately. The fourth fraction contained the liposomes and was used for the CF-release study.


93

A sodium chloride solution of 70 mM (135 mOsm) was used to match the osmolality of the

entrapped CF solution (129 mOsm measured with the advanced MICRO-OSMOMETER, Vitech

Scientific Ltd, West Sussex, UK) which should not differ significantly from that of the sodium chloride solution outside of the liposomes (234).

4.3.3.2 5(6)-carboxyfluorescein (CF) release profile

Fluorescence spectroscopy was used to evaluate the ability of chlorhexidine digluconate (CHD)

and PP-A148 to disrupt the bilayers of vesicles composed of E. coli lipids. In each experiment 50 μL of CF-loaded liposomes were added to 2350 µL of NaCl (70 mM) in a cuvette (10 × 10 mm)

and the fluorescence intensity was measured using a Varian Cary Eclipse fluorimeter (Agilent,

USA) at 25 °C. Antimicrobial solutions (3.15 mM) were added (0.1 ml) after 60 s and the

fluorescence intensity of the mixture was again measured at an excitation/emission wavelength

of 490/510 nm and a slit width of 2.5 nm for both processes. The fluorescence was monitored at

hourly intervals, and at the end of the experiment the total concentration of CF was determined

after lysing the liposomes with 100 μL of Triton X-100 (10 w/v%). CF-liposomes in sodium chloride (70 mM) were also monitored hourly, to determine their integrity. The data was

normalized against controls designed to compensate for the effects of dye photobleaching over

the duration of the experiment. The percentage dye release was calculated using the following

formula:

𝐶𝐹% = ¡{�¡z¡��¡z

× 100, 𝑓𝑜𝑟𝑡 ∈ [0, 𝑇] Equation 4.3

, where 𝐹L is the fluorescence intensity of each sample at time t, 𝐹w is the fluorescence intensity

prior to the addition of the drug (liposomes in NaCl) of each time point, and 𝐹� is the intensity of

each sample after the addition of Triton X-100, which corresponds to 100% dye release.

4.3.4 Neutron reflectivity of asymmetric bilayer of DPPC and LPS and interaction

with fluidosomes

4.3.4.1 Asymmetric bilayer deposition

The lipid components of model Gram negative bacterial membranes were deposited on a

piranha-cleaned (SiO2) surface of single silicon crystals (50×80×20 mm), using a Langmuir-

Blodget (LB) trough (KSV-NIMA, Biolin Scientific, Finland) (Figure 4.1). The LB technique was

used to deposit the inner leaflet of the membrane (d62DPPC) on the 50×80 mm polished surface of the crystal, and the Langmuir-Schaefer (LS) deposition used for the outer leaflet of the

membrane (Ra LPS) (Figure 4.1). Three isotherm cycles were conducted prior to deposition to

examine the stability of the monolayers. For the LB deposition, the silicon block was submerged

into the ultrapure non-buffered water subphase (containing 5 mM CaCl2), cooled to 10°C. A

solution of d62DPPC in chloroform was deposited dropwise (50 uL of 2.5 mg/mL) on the subphase

surface and compressed slowly to a constant surface pressure of 38 mN/m. The block was lifted


94

through the air-water interface at a speed of 4 mm/min at a constant surface pressure at 38

mN/m. The LB trough was then cleaned and the air-liquid interfacial monolayer of Ra LPS (2.5

mg/mL) was deposited (from 60% chloroform, 39% methanol and 1% water v/v) on to the same clean non-buffered interface, cooled at 10°C, with a surface pressure of 38 mN/m. The LS

deposition was achieved by placing the silicon block covered with a homogeneous d-DPPC

monolayer (with the headgroup to the silicon block and the hydrophobic tails facing the

environment) to a holder above the water surface with the alcyl chains pointed downward, toward

the interface. The surface of the crystal was made parallel with the interface by adjusting the

angle of the crystal using a built-in automatic levelling device. The silicon block was lowered at

a speed of 3 mm/min through the interface in order to allow the lipid monolayer on the block to match the LPS monolayer on the subphase via the hydrophobic chains, producing a bilayer. The

block was then allowed to continue moving downwards until the bottom of the trough was reached

where a customised PTFE sample cell was placed. The sample cell and silicon block were placed

in a custom-made metal holder to ensure the bilayer was fully contained and sealed inside the

small water chamber of the sample cell, whose total volume was 3 mL.

Figure 4.1: Langmuir-Blodget and Langmuir-Schafer deposition on a silicon crystal to manufacture the assymetric model outer membrane of Gram negative bacteria.

4.3.4.2 Neutron reflectometry measurements and data analysis

Specular neutron reflectometry (NR) measurements were carried out using the INTER

reflectometer at the ISIS neutron source, Rutherford Appleton Laboratory (Oxfordshire, UK),

using neutron wavelengths from 1 to 16 Å. The reflected intensity was measured at two incident

angles of 0.7° and 2.3° as a function of the momentum transfer Q (𝑄 = OA PQRSB

, where λ is the


95

wavelength and θ is the incident angle). The purpose-built flow cells of the silicon-liquid interface

were placed on an anti-vibration sample stage and the inlet of the cell was connected to a L7100

HPLC pump (Merck, Hitachi, Germany). The pump allowed the easy exchange of the isotopic contrast solutions within the cell (3 mL) at a flow rate of 1.5 mL/min. The membrane was

measured at a temperature of 25°C and then at 38°C but the challenge with liposomes was

measured only at 38°C. Temperature was kept constant throughout the experiment by using a

circulating water bath. Empty fluidosomes were manufactured following the thin film hydration

method as described in Chapter 3 (section 3.3.1). Lipid film was re-dispersed using ultrapure

water and fluidosomes (0.1 mg/mL) were injected into the sample cell and allowed to equilibrate

for 1 h. The excess fluidosomes were flushed out of the cell chamber before data acquisition so as to measure the possible changes caused by their interaction with the model OM.

Each sample was examined under three different isotopic contrast conditions, i.e. 100% H2O,

100% D2O and silicon matched water or SMW (38% D2O, 62% H2O), all containing CaCl2, to

highlight the different components of the bilayer structure.

4.3.4.3 NR data and Statistical analysis

The three contrasts produced three reflectivity profiles which were simultaneously fitted using

models describing the interfacial structure based on Adele’s, and Born and Wolf’s matrix

formulism using the RaScal software (231,232). In this approach, the interface is described as a

series of slabs and the software fits a layered model to the structure by calculating common parameters, i.e. scattering length density (SLD), thickness, roughness and hydration state of

each layer.

Table 4.1: Summary of scattering length densities of the lipid components studied, and the solution subphases.

Lipid/Solvent Neutron Scattering Length Density (SLD)/(× 10-6 Å-2)

Silicon 2.07 Silicon Oxide 3.41 DPPC headgroup 1.98 d-DPPC tails 7.45 Ra-LPS headgroup (core oligosaccharide)/D2O 4.28

Ra-LPS headgroup (core oligosaccharide)/H2O 2.01

For the model membrane consisting of chain-deuterated d62DPPC and hydrogenous rough Ra-

LPS (E. coli, EH100), the calculated scattering length density (Table 4.1) and the fitted

parameters were used to examine the asymmetry of the bilayer. The lipid coverage in each leaflet

was calculated from the volume fraction of bilayer defects across the surface by calculating the

water volume fraction and the deuterated PC tail fraction (225). The water volume fraction can


96

be determined by the difference in SLD of the DPPC and LPS tails in the three solvent contrasts.

In the tail region, both lipid tails do not possess labile hydrogens and, therefore, do not undergo

changes in SLD (ρ). The mathematical model uses the following formulas to calculate the water volume fraction from the raw data, and the parameter is called “Tail-hydration”.

𝜌FHLLuW = 𝜌b¤¤a𝜑b¤¤a + 𝜌�¤v𝜑�¤v + 𝜌�@¦𝜑�@¦ Equation 4.4

𝜑�@¦ =YTUU§¨{gT©ª«U{¨gh{¬�YTUU§¨{gT©ª«U{¨gh{Z

Yª«U{¨gh{¬�Yª«U{¨gh{Z Equation 4.5

where 𝜌FHLLuW is the fitted SLD value of the bilayer, 𝜌b¤¤a , 𝜌�¤v, 𝜌�@¦ are the calculated SLD of

the deuterated DPPC tails, hydrogenous LPS tails and water, respectively, 𝜑b¤¤a , 𝜑�¤v and

𝜑�@¦ are the volume fractions, 𝜌HIIuGLkH�oJILGktLV, 𝜌HIIuGLkH�oJILGktL@, 𝜌oJILGktL and are the

experimental SLD values of the inner tails in two contrast solutions (for example H2O and D2O)

and the experimental SLD of the two chosen contrast solvents. Based on Equation 4.4 and

Equation 4.5, the volume fraction of deuterated DPPC tails was calculated using the following

formula:

𝜑b¤¤aLkH�t =Y�YZ®¯Z®�¯�°±{gT©h(V�¯Z®)

Y²³°°´�Y�°±{gT©h Equation 4.6

, where 𝜌 is the fitted SLD, 𝜌�@¦,𝜌Wb¤¤a and 𝜑�¤vLkH�t is the SLD of the water, DPPC and LPS

are the experimental values. The PC-tail volume fraction 𝜑b¤¤aLkH�t was calculated before taking

into account the hydration of the bilayer tail region, which is a separate fitted parameter of the

mathematical model. Therefore, the volume fraction of the LPS tails 𝜑�¤vLkH�t was reduced by the

following formula:

𝜑�¤vLkH�t = 1 − (𝜑b¤¤aLkH�t + 𝜑�@¦) Equation 4.7

All three samples (RT, 38°C before and after liposome introduction) shared the same silicon

substrate because the same model membrane was measured in different conditions. The models were fitted to the data using a Bayesian Markov Chain Monte Carlo (MCMC) algorithm (227,233).

In addition to the model parameters, the backgrounds, scale factors and instrument resolutions

were also fitted. A Bayesian approach was followed for the fitting, with prior probability distribution

(priors) chosen according to already known information prior to the analysis, posteriors obtained

using a Delayed Rejection (DR) algorithm and the best fit parameters taken as the distribution

maxima. DR is a way of modifying the standard Metropolis-Hastings (MH) algorithm to improve

the efficiency of the resulting MCMC estimators and decrease the asymptotic variance (234).

The idea is that upon rejection in an MH algorithm, instead of advancing the run time and retaining the same position, a second stage is proposed with improved distribution. This process

can be iterated for a fixed or a random number of stages, and higher stage proposals depend on

the proposed and rejected fits. The error analysis was run at 95% confidence intervals for each

distribution.


97

4.4 Results

4.4.1 Surface Pressure-Area isotherms of Langmuir monolayers at the air-liquid

interface

Pressure-area isotherms were used to determine the physical state and mechanical properties

of the E. coli polar lipid extract, DPPC/DPPG [3/1], POPC/POPG [3/1], Rc LPS J5, Ra LPS

EH100 and Lipid A monolayers. E. coli B lipid extract and both PC/PG monolayers were used to

mimic the Gram negative IM. Rc J5 LPS J5 and R595 Lipid A, both of which were used to model

the outer leaflet of the OM. The average molecular weight of each mixture was used to plot the

surface pressure-area isotherms (Figure 4.2) which showed the phase transitions of each monolayer during compression. Changes in the slope and shape of the curves during

compression resulted from changes in the orientation, packing and arrangement of phospholipids

at the interface.

4.4.1.1 Inner membrane model monolayers

All pressure-area isotherms in (Figure 4.2A, C, E) are representative curves chosen from

triplicate measurements for each lipid monolayer. The isotherm obtained from pure DPPC

monolayers (Figure 4.2C) nearly reached the solid state because the slope was almost

perpendicular to the y-axis (235,236), whereas from E. coli B, POPC, POPG and POPC:POPG

(3:1) (Figure 4.2A, E) displayed typical expanded monolayers.

DPPC (hydrogenous and deuterated) and DPPC/DPPG undergo an L1/L2 transition, marked by

a plateau (or pseudo plateau) in the isotherm and a corresponding minimum in the compression

modulus plot (Figure 4.2C and D). The transition began at approximately 4 mN/m for h-DPPC

and the mixture (237,238), while for d-DPPC it started at 9mN/m (239). The DPPC/DPPG

monolayer reached its liquid condensed phase just below 20 mN/m with a maximum Es of 197

mN/m. For POPC/POPG, the transition to liquid condensed state was initiated around 25 mN/m,

making it less compressible than the DPPC/DPPG monolayer (Figure 4.2E, F). The E. coli B

display areas of liquid-expanded phases before turning into the coexistence of L1/L2 phase at surface areas of ~ 15 mN/m, and remained until the collapse of the monolayer (Figure 4.2A, B).


98

Figure 4.2: A, C and E present the Surface Pressure-Area (P-A) isotherms generated at 23 °C for a series of pure or mixtures of phospholipids (E. coli extract, DPPC, DPPG, POPC, POPG). All phospholipids were suspended in chloroform apart from d62DPPC which was prepared in chloroform:methanol:water (6:4:1) before deposition. The subphase used was 1mM MgCl2 for all monolayers apart from d62DPPC which was 5 mM CaCl2. The E. coli B monolayer was deposited either on ultra pure water or 0.9% NaCl The subphase for hydrogenous DPPC/DPPG and POPC/POP mixtures monolayers was MgCl2. E. coli isotherm was measured on ultra pure water and isotonic saline. The red arrows show the L1/L2 and the L2/S transition of the lipids where applicable. B, D and F show the calculated compressibility modulus derived from P-A isotherms at 23 °C. The red arrows show transition phases of the lipids which are more pronounced than in the P-A isotherm.


99

Figure 4.3: Collapse surface pressure (A), area per molecule (B) and compressibility modulus (C) at 30 mN/m of all the phospholipid lipid monolayers

Figure 4.3B presents the area per molecule at 30 mN/m, which is the equivalent lateral pressure

of the bacterial inner membrane (130), and the collapse pressure of each lipid monolayer (3

Figure 4.3C). E. coli B present an average area per molecule of 52.0 ± 4.0 Å2 in water and 62.6

± 9.1 Å2 in the presence of 0.9% NaCl, which was not statistically different from that in water at

a confidence interval of 95% (P=0.6). The collapse pressure of the monolayer is very low

considering that 30 mN/m is the equivalent pressure of a bacterial membrane (35.4 ± 5.6 and

31.1 ± 0.1 mN/m in water and saline respectively) (130). This monolayer caused overflow at the

edges of the trough, probably due to the PE shape and orientation.

The structural difference between POPC (C16:0-18:1) and DPPC (16:0) affects their packing and

orientation. The average molecular area of POPC, POPG was 58.0 ± 0.9 Å2 and 62.7 ± 0.5 Å2,

respectively, with slightly higher collapse pressure (Figure 4.3A). The average area per molecule

(A) of an ideal mixture was calculated by the additivity relationship considering the molar fraction

X of each component (244):


100

𝐴 = 𝑋V𝐴V + 𝑋@𝐴@ Equation 4.8

Based on this rule within a mixed monolayer and the experimental values of the single

components (POPC and POPG), the resultant area per molecule would be 59 Å2, which in

agreement with the experimental value (60.2 ± 0.9 Å2), assuming an ideal mixing of lipids. DPPC

based monolayers (h- and d-) exhibited similar area per molecule (approximately 40 Å2) at 30

mN/m. Based on the experimental molecular area of h-DPPC from this experiment and that of DPPG published in the literature (40 Å2) (240), the mixed monolayer should have an average

molecular area of 42 Å2, which is in favorable agreement with the experimental value (44.72 Å2).

The mixture DPPC:DPPG display high collapse pressure at 51.4 ± 0.5 mN/m, while h-DPPC

collapses at 41.4 ± 3.9 mN/m, both of which lie within the normal collapse pressure range

(235,241).

4.4.1.2 Outer membrane model monolayers

The initial suspension of Rc J5 LPS, Ra EH100 LPS and R595 Lipid A were used to model the

outer membrane under the same experimental procedure. Rc-LPS J5 and Ra EH100 LPS

presented a typical expanded isotherm because their compressibility modulus did not exceed 50

mN/m with increasing surface pressure (Figure 4.4). Two plateaus (or pseudo-plateaus) were

observed for Rc J5 LPS at a surface pressure of approximately 4 and 25 mN/m, which reflect changes in the lipid packing, while remaining in the L1 phase (Figure 4.4). Re Lipid A from R595

showed a coexistence of L1 and L2 at around 15 mN/m, which changed into a more ordered and

tightly bound state upon further compression (242–244). A sudden drop in the surface

compressibility modulus profile indicates the collapse of the monolayers, which is pronounced

for Lipid A (Figure 4.4B).

Figure 4.4: (A) Surface Pressure-Area (P-A) isotherms generated at 23 °C for a series of glycolipids: Ra EH100 LPS, Rc J5 LPS and Lipid A. The subphase for Rc-LPS and Lipid A monolayers was MgCl2 while CaCl2 was used for Ra-LPS. The red arrows show the L1/L2 and the L2/S transition of the lipids where applicable. (B) Compressibility modulus derived from P-A isotherms at 23 °C. The red arrows show transition phases of the lipids which are more pronounced than in the P-A isotherm.


101

The onset of pressure increase upon compression (Figure 4.4A), also known as the lift-off area,

appear over a much larger area per molecule for the Ra-LPS (400 Å2) than the corresponding

value for the Re-Lipid A (200 Å2) and Rc-LPS (269 Å2) monolayers, which can be explained by the bulkier head group of LPS Ra, followed by Rc-LPS, and lastly Re-Lipid A.

Figure 4.5: Collapse surface pressure (A), area per molecule (B) and compressibility modulus (C) at 30 mN/m of three types of glycolipid monolayers. The Ra-LPS did not reach its collapse pressure and it is excluded from the figure.

Rc-J5 LPS exhibited a molecular area of 102.0 ± 2.6 Å2 at 30 mN/m at a surface pressure of 30

mN/m with a high collapse pressure (45.0 ± 0.5 Å2) while the Ra-EH100 LPS presented larger

area (164.0 ± 43.0 Å2) but did not reach the collapse point (Figure 4.4A, Figure 4.5A, B). Re

Lipid A had a molecular area of 95.66 Å2 at 30 mN/m and collapse pressure of 44.1 which

conform with the results of Garcia-Verdugo et al. and Jeworrek et al. (219,242). They used a

different subphase solution (150 mM NaCl, 5 mM Tris-HCl pH 7.4, 150 μM CaCl2) which resulted

in a more expanded headgroup with a size of approximately 300 Å2.

4.4.2 Drug interactions with model inner membranes

The interaction studies of the monolayers and bilayer with antimicrobial agents and liposomes

were carried out at 30-35 mN/m. A comparison of the compressibility modulus at lateral bacterial

membrane pressure (Figure 4.3C and Figure 4.5C) revealed differences between the

monolayers in terms of orientation and packing, which may have affected the drug-membrane interaction. DPPC/DPPG (Es = 151.8 ± 18.7 mN/m) exhibited a more ordered and closely packed

monolayer than POPC:POPG (Es = 104.2 ± 7.9 mN/m). E. coli B monolayer displayed a liquid

expanded state. As for the OM model monolayers, Lipid A (Es = 99.9 ± 16.9 mN/m) presented a

liquid condensed state at 30 mN/m compared to Rc-LPS (51.4 ± 4.3 mN/m) which remained fluid

and less ordered.

The Langmuir monolayer interfacial technique was applied to examine the interaction of drug

and cyclodextrin with the lipid monolayers representing the IM and the outer leaflet of the OM.

The lipid monolayers were formed on an aqueous subphase at 23°C and a PFA petri dish (50

mm diameter, 20 mL volume capacity) was used to spread the lipids up to the desired surface

pressure (30-35 mN/m) before the antibiotic and cyclodextrin solutions were injected below the


102

surface. The surface pressure was plotted against time and the curve was fitted to a non-linear

mathematical model (Hiil equation), where applicable, to examine the kinetics of the interaction.

The parameters gave an insight into the affinity of the studied molecules towards the monolayers (Hill-slope), the rate of change in surface pressure (t50%), and the efficiency of the interaction in

terms of the maximum change in surface pressure (ΔΠmax).

Chlorhexidine, rifampicin and gentamicin were used as reference drugs to judge the efficiency of

the interaction and investigate the uptake mechanism for the novel antimicrobial agent, PPA148.

Chlorhexidine is known as a membrane active drug which breaches the cell envelope and causes leakage of cellular content leading to bacterial death (250). Rifampicin is transported into the cell

membrane (OM and IM) by passive diffusion, while gentamicin follows a self-promoted pathway

through the OM and needs energy input for its transfer inside the IM by specific porins

(32,33,122). The controls will be compared with PPA148’s interaction profile to investigate its

passive transport and whether the novel drug disrupts the model membranes or diffuses through

them.

4.4.2.1 Interaction with E. coli polar lipid monolayer

E. coli B is a natural lipid extract which was used to model the Gram negative inner membrane.

The monolayer presented an liquid expanded phase at 30 mN/m, as found from its isotherm and

compressibility profile (Figure 4.3C). The surface pressure-time isotherm clearly shows the

disruption effect of CHD on the monolayer (Figure 4.6A). CHD penetrated the monolayer and adsorbed at the air-water interface causing a large increase in surface pressure (ΔΠmax = 11.9 ±

1.7 mN/m) within seconds (Figure 4.6A).

The interaction between the novel antibiotic PPA148 and the E. coli B monolayer was examined

by increasing drug concentrations (Figure 4.6C, Table 4.2). The isotherms of 20 and 60 μg/mL

(Figure 4.6C) were fitted with the Hill equation to investigate the kinetics of the interaction. PPA148 at 2 μg/mL presented a negative change in surface pressure, possibly indicative of a

propensity to cause membrane damage within 8 hours. By increasing PPA148’s concentration,

the interaction profile showed a positive change in surface pressure with ΔΠmax of 14.0 ± 3.8

mN/m and 9.2 ± 2.9 mN/m for 20 and 60 μg/mL (Figure 4.6C, Table 4.2). ΔΠmax was reached

within 4 hours and the drug’s affinity towards the lipid monolayer was higher at 20 μg/mL PPA148.

The magnitude of ΔΠmax might indicate that PPA148 penetrated the monolayer. Due to poor

water solubility of PPA148 (30 ± 1 μg/mL as ascertained in Chapter 2), drug precipitation was

observed after injecting 60 μg/mL into the subphase, possibly explaining the weaker affinity (ΔΠmax) of the drug towards the E. coli model membrane.


103

Figure 4.6: Representative surface pressure – time isotherms at 23°C of chlorhexidine digluconate (A), rifampicin (B), gentamicin (B), PPA148 (C) and DMSO (0.5, 1 and 1.5 % v/v) (D) against E. coli B monolayer using 0.9% NaCl as a subphase. The black lines are the fitted curves based on the mathematical model, Hill curve, when it was applicable.

Table 4.2: Kinetic parameters obtained from the fitting of the binding isotherms of rifampicin, gentamicin and PPA148 and the E. coli B monolayers at 23°C on water sub-phase containing isotonic saline.

Drug Concentration (μg/mL)

ΔΠmax (mN/m) Hill slope

t50% (sec or h)

CHD 2 11.9 ± 1.7 4.6 ± 2.4 19.8 ± 5.8 sec

Rifampicin 2 -3.7 ± 1.6 N/A 3.6 ± 1.1 h Gentamicin 2 -6.6 ± 1.5 N/A 2.4 ± 1.1 h

PPA148 2 -5.1 ± 2.7 N/A 3.7 ± 1.3 h PPA148 20 14.0 ± 3.8 1.2 ± 0.5 2.8 ± 2.2 h PPA148 60 9.2 ± 2.9 0.8 ± 0.1 1.3 ± 0.3 h

The injecting solution for PPA148 was 100% DMSO, which is known to affect lipid membranes. Therefore, aliquots of DMSO in 0.5%, 1% and 1.5% v/v (0.001, 0.003 and 0.005 mole fraction of

DMSO in the trough) were injected into the subphase below the monolayer (Figure 4.6D). The

fractions tested correspond to the injected drug sample volumes (0.1, 0.2 and 0.3 mL). Figure

4.6D shows that DMSO at 0.5 or 1% v/v elicit a maximum change in surface pressure of 2 mN/m

or less indicating a very weak or zero interaction of DMSO molecules with the lipid headgroup.


104

However, at the highest concentration tested (1.5 % v/v), the interaction with lipid molecules was

more pronounced, reaching ΔΠmax of 7.0 ± 1.3 mN/m.

The profile of the two other control drugs (gentamicin and rifampicin) (Figure 4.6B) are

incompatible with their established mechanisms of uptake. Gentamicin sulfate, which was used

as a negative control, interacted with E. coli monolayer by inducing a negative change in surface

pressure, thus appearing to solubilize the monolayer (Figure 4.6B). However, rifampicin, the

positive control, did not show any sign of interaction. These results suggest this planar monolayer

is an unsuitable model for the E. coli B lipids. The interaction of non-membrane active compounds might be artefactual due to the monolayer’s packing inconsistencies and loosely packed lipids

on a planar surface. Therefore, in the next section, fluorescence spectroscopy was used to

examine whether PPA148 can really damage a membrane, by measuring the release of a dye

from E. coli B liposomes.

4.4.2.2 Interaction of PPA148 with E. coli B liposomes using fluorescence spectroscopy

The interaction experiments with E. coli B monolayer was used as an indication as to whether

PPA148 might passively diffuse into the model membrane. The monolayer experiment showed

possible damage of the E. coli B membrane caused by PPA148. However, the results were

inconsistent for rifampicin and gentamicin, and thus further experiments were carried out to gain

a better understanding of the interaction. Challenging liposomes with drugs and monitoring the

release of carboxyfluorescein illustrates whether a drug can disrupt a membrane. Lipids from the E. coli B extract tend to form non lamellar structures and unstable planar monolayers, due to the

presence of PE and its negative curvature (Figure 1.3). These lipids are likely to form a more

stable configuration when are organized in liposomes. The dye self-quenches inside the

liposomes; but when it is released into the bulk solvent it is no longer quenched, resulting in a

significant increase in fluorescence intensity (245). Therefore, increase in fluorescence intensity

indicates a release of the dye by disruption of the lipid bilayer caused by the antimicrobial agents

tested.

Figure 4.7: CF release from liposomes, manufactured from natural E. coli lipid extract, after administration of PPA148, chlorhexidine digluconate and gentamicin at 25°C.


105

In this experiment chlorhexidine digluconate (CHD) and gentamicin were used as positive and

negative control respectively, since the former is known to be a membrane active compound and

to damage membranes, while the latter enters the bacterial cells via active transport. The partitioning of CHD proved to be disruptive to the E. coli bilayer since the drug caused 65.7 ±

9.5% dye release, whereas the detergent effect of PPA148 apparent in the Langmuir isotherm

(Figure 4.6C) was not confirmed by the dye release study (Figure 4.7). The novel antimicrobial

only elicited a limited degree of dye release (4.2 ± 2.2%) one hour after the addition of the drug,

without any further increase over the next 3 hours. This profile does not describe the solubilizing

effect observed using E. coli B monolayer. This fact is evidence that the E. coli B monolayer is

not a good model to study interactions, despite being extracted from bacteria. Therefore, the other interaction experiments using model monolayers at the air/liquid interface were carried out

using only DPPC/DPPG lipids, which form a more stable planar monolayer.

4.4.2.3 Interaction with DPPC:DPPG monolayer

A mixture of DPPC/DPPG lipids was used as an alternative IM model to the E. coli B monolayer. Rifampicin at 2 μg/mL interacted slowly (t50% = 4.8 ± 0.9 h) with the DPPC:DPPG monolayer by

reaching a maximum pressure of 6.7 ± 2.0 mN/m after 10 hours (Figure 4.8A). By increasing the

drug concentration from 2 to 20 μg/mL, the Hill slope decreased from 2.5 ± 1.0 to 0.8 ± 0.2 and

ΔΠmax increased to 15.8 ± 5.3 mN/m, indicating a faster and more efficient interaction (Figure

4.8A, Table 4.3). Rifampicin at 30 μg/mL induced ΔΠmax of 3.4 ± 0.6 and t50% of 1.1 ± 0.5 h,

reflecting a weaker and slower binding capacity.

Figure 4.8: Representative surface pressure – time isotherms at 23°C of rifampicin (A), PPA148 (B), against DPPC/DPPG monolayer using 1 mM MgCl2 as a subphase. The black lines are the fitted curves based on the mathematical model when it was applicable.

PPA148 at 2 μg/mL gave rise to a ΔΠmax of 8.8 ± 3.4 mN/m within 10 hours with a Hill slope of 1.7 ± 0.3 and t50% of 5.5 ± 1.54 h, reflecting an efficient but slow interaction with the DPPC/DPPG


106

monolayer (Figure 4.8B). At higher drug concentrations, the slope of the isotherm was steeper,

which indicates an increased affinity towards the lipid monolayer. At 20 μg/mL, the extent of the

surface pressure modification was lower and ΔΠmax reaches only 3.3 ± 0.7 mN/m, compared to the ΔΠmax obtained with 2, 40 and 60 μg/mL which were, respectively, 8.8 ± 3.4, 9.4 ± 1.3 and

6.4 ± 1.4 mN/m (Table 4.3). This might have been a result of drug aggregation, which started at

~20 μg/mL and visible aggregates was seen above 30 μg/mL, as found in the turbidimetric assay

(Figure 2.15) in Chapter 2 detected by dynamic light scattering.

Overall, the rate of interaction at concentrations above 2 μg/mL indicated a faster process, as observed from the t50%. At concentrations above 20 μg/mL, drug precipitation was observed after

injecting PPA148 in the trough which agrees with the water solubility (30 ± 1 μg/mL) found in

Chapter 2, Table 2.3. Overall, the affinity and kinetics of the interaction between PPA148 and

DPPC/DPPG is concentration-dependent, with the kinetics having a threshold of 20 μg/mL.

Table 4.3: Kinetic parameters obtained from the fitting of the binding isotherms of rifampicin and PPA148 and the DPPC/DPPG monolayers at 23°C on water sub-phase containing1 mM MgCl2.

Drug Concentration (μg/mL)

ΔΠmax (mN/m) Hill slope t50%

(h)

PPA148

2 8.8 ± 3.4 1.7 ± 0.3 5.5 ± 1.54 20 3.3 ± 0.7 1.2 ± 0.7 0.7 ± 0.4 40 9.4 ± 1.3 1.1 ± 0.4 0.9 ± 0.2 60 6.4 ± 1.4 0.6 ± 0.0 1.3 ± 0.2

Rifampicin 2 6.7 ± 2.0 2.5 ± 1.0 4.8 ± 0.9 20 15.8 ± 5.3 0.8 ± 0.2 3.2 ± 2.4 30 3.4 ± 0.6 1.2 ± 0.6 1.1 ± 0.5

4.4.2.4 Interaction between cyclodextrins and IM model membranes

One of the hypotheses of this project is that cyclodextrin will enhance not only PPA148’s water

solubility but also its transport through the inner bacterial membrane. Therefore, the effect of

HPβCD and RAMEB on the IM was tested using phospholipid model monolayers. Both types of cyclodextrins showed a solubilization effect on all lipid monolayers by eliciting a negative change

in surface pressure. DPPC/DPPG (3:1) monolayers underwent a significant loss of molecules

from the surface when challenged with HPβCD and RAMEB (Figure 4.9A, B). However, RAMEB

had a stronger effect against POPC/POPG than HPβCD by producing a greater drop in surface

pressure. The curves therefore indicate damage to the phospholipid monolayer rather than

partitioning of the cyclodextrins.


107

Figure 4.9: Representative surface pressure – time isotherms at 23°C of HPβCD (A) and RAMEB (B), against DPPC/DPPG (3:1) and POPC/POPG (3:1) monolayers using MgCl2 as a subphase.

4.4.3 Drug interaction with model outer membranes

4.4.3.1 Interaction between PPA148, fluidosomes, and model OM monolayers

Lipid A and Rc-LPS J5 monolayers were used to investigate the effect of steric barrier from the

OM on PPA148, rifampicin and fluidosomes. Lipid A constitutes the innermost compartment of LPS and lacks the polysaccharide chain region. In this experiment rifampicin was used as a

positive control because it shows a very good correlation with PPA148 when tested against

model IM lipid monolayers (Figure 4.8). Rifampicin (20 μg/mL) did not affect the packing of the

Lipid A monolayer and the surface pressure remained constant for 2 h (Figure 4.10B). Instead,

PPA148 produced an increase in surface pressure of 6.9 ± 0.7 mN/m, reaching a plateau in less

than an hour with a high binding affinity (Hill slope = 1.3 ± 0.2) (Figure 4.10A).

When PPA148 was used to challenge Rc-LPS J5 monolayers, it revealed low efficiency (ΔΠmax

= 3.7 ± 2.1 mN/m), weak binding capacity (Hill slope = 3.7 ± 2.6) but a fast rate of t50% = 0.2 ± 0.1

h (Figure 4.10A). The presence of the polysaccharide chain hindered the interaction of PPA148

with the J5 LPS monolayer, probably due to a steric effect. The affinity (Hill slope) and efficiency

(ΔΠmax) of PPA148 towards J5 LPS was non-reproducible (Table 4.4).


108

Figure 4.10: Interaction isotherm of 20 μg/mL PPA148 (A), 20 μg/mL rifampicin (B) and 0.1 mg/mL fluidosomes (C) against different types of OM lipid monolayers (Lipid A and Rc-LPS J5) at constant surface area at 23 °C. The profile was fitted, when applicable, to specific binding Hill slope model to investigate the kinetics and affinity of the interaction. The black lines are the fitted curves based on the mathematical model.

Fluidosomes were tested against J5 LPS monolayer and showed a linear decrease in surface

pressure (Figure 4.10C). A ΔΠmax of 2 ± 0.4 mN/m was reached 3 h after injection of the final

formulation below the monolayer, without any sign of reaching a plateau, suggesting removal of

the lipids from the surface.

Table 4.4: Kinetic parameters obtained from fitting the binding isotherms of rifampicin and PPA148 and the Rc J5 LPS and R595 Lipid A monolayers at 23°C on a water sub-phase containing1 mM MgCl2.

Drug Parameters Rc J5 LPS Re R959 Lipid A

PPA148

ΔΠmax (mN/m) 3.7 ± 2.1 7.9 ± 0.7

Hill slope 3.7 ± 2.6 1.3 ± 0.2

t50% (h) 0.2 ± 0.1 0.3 ± 0.1

Rifampicin ΔΠmax (mN/m) - 0.0 ± 0.0

Hill slope - N/A t50% (h) - N/A


109

4.4.3.2 Interaction between cyclodextrins and model OM monolayers

Although cyclodextrins were primarily used in this study to enhance PPA148’s water solubility (Figure 2.17, Table 2.3) and help its transport through the IM, they were also examined for their

interaction with OM LPS. Lipid A and Rc-LPS J5 remained largely undisturbed by HPβCD, as

reflected by the minimal surface pressure change (within ± 2 mN/m), whereas RAMEB induced

a weak interaction with a more important change in surface pressure (approximately -5% mN/m)

(Figure 4.11). These findings show that there was no significant interaction between the two

cyclodextrins and the OM model membranes.

Figure 4.11: Pressure-time interaction isotherm of HPβCD (A) and RAMEB (B) with 2 monolayers under constant surface area at 23°C.

4.4.3.3 Interaction between fluidosomes and model OM using neutron reflectivity (NR)

NR was used to further explore the effect of fluidosomes on a model bilayer representing the

Gram negative OM. The structure of the model membrane was characterized at room

temperature and at 38 °C prior to being challenged with the liposomal carrier (hydrogenated

DPPC/DMPG at a molar ratio of 18:1) at 38 °C. Each contrast variation (H2O, D2O and SMW)

was fitted by mathematical models, describing the various layers of the membrane. The interface

has been described as a series of slabs, each characterized by its SLD, thickness, hydration and roughness. This model membrane has been extensively studied by other research groups and

the preparation process gives robust results with minor differences among samples

(135,225,226).

Figure 4.12 shows the reflectivity and SLD profile of the model outer membrane at room

temperature. The NR data were fitted to a five-layer model, thereby seeing the minimal number of layers with which the reflectivity data could be fitted. The interference fringe observed in the

D2O contrast (Figure 4.12A) reflects the presence of the bilayer. The layers, presented as a


110

background to the fitted SLD curves, are the silicon, silicon oxide, inner d-DPPC headgroups,

inner d-DPPC tails, outer Ra-LPS tails, Ra-LPS core region (headgroup) and bulk solution.

Figure 4.12: Neutron reflectivity profile (A) and model data fits with their scattering length density profiles (B) for an assymetrically deposited DPPC (inner leaflet) and Ra-LPS (outer leaflet) model membrane at room temperature. The sample was measured at three isotopic contrasts (100% D2O, 100% H2O and SMW). The model membrane was fitted into five-layered mathematical model: Silicon oxide (SiO2), DPPC headgroup (Inner HG), DPPC tails (Inner tails), Ra LPS tails (Outer tails) and Ra LPS headgroups (Core).

The results reveal that a highly asymmetric lipid bilayer was formed on the silicon oxide-coated

surface of the substrate with an average lipid coverage of 99%, derived from the combined

volume fractions of d-DPPC and LPS (Table 4.5). The results have shown a mixing of DPPC and

LPS within the bilayer because the DPPC composition in the inner and outer leaflet was found to be 77% and 20% respectively (Table 4.5). This is in good agreement with published values

(226). A total of 1.3% water was found within the tail layers with a bilayer roughness of 4.15 Å.

(Table 4.6)

Table 4.5: Volume fractions of deuterated DPPC tails, hydrogenous LPS tails and water within the bilayers of Ra-LPS/d-DPPC at room temperature (25 °C).

Layer 𝝋𝑫𝑷𝑷𝑪 𝝋𝑹𝒂𝑳𝑷𝑺 𝝋𝒘𝒂𝒕𝒆𝒓 SiO2 n/a n/a 0.16

(0.15, 0.17) Inner Headgroup 0.77

(0.75, 0.78)

0.22 (0.21, 0.24) 0.013

(0.012, 0.013)

Inner Tails

Outer Tails 0.20 (0.19, 0.21)

0.79 (0.77, 0.80) Outer

Headgroup/Core


111

At 38 °C the membrane displayed a similar reflectivity profile and the data were fitted using the

same five-layer model (Figure 4.13). However, the structural parameters changed slightly (Table

4.6; Table 4.7). The d-DPPC volume fraction did not change in the inner leaflet (0.78) whilst increasing (0.14) in the outer leaflet. The thickness of the inner and outer headgroups decreased

from 7.66 Å and 28.35 Å at 25 °C to 7.41 Å and 23.06 Å at 38 °C respectively. Moreover, the

hydration of the inner headgroup decreased from ~50 % to ~28 ± 1 % and the bilayer roughness

increased by ~ 2 Å, which was expected due to the increase in fluidity of the bilayer.

Table 4.6: Structural parameters, thickness, hydration and roughness, of the Ra-LPS/DPPC membrane at room temperature and 38 °C as obtained from the fitting of the neutron reflectivity data. The numbers in parentheses are the 95% confidence interval error.

Layer

Asymmetric bilayer at:

RT 38 °C

Thickness (Å)

% water

Roughness (Å)

Thickness (Å)

% water

Roughness (Å)

SiO2 20.13

(19.58, 20.72)

16.27 (15.56, 16.90)

4.61 (4.39, 4.86)

20.13 (19.58, 20.72)

16.27 (15.56, 16.90)

4.61 (4.39, 4.86)

Inner HG 7.66

(7.346, 7.98)

50.10 (47.86, 52.38)

4.15 (4.02, 4.30)

7.41 (7.03, 7.80)

28.27 (27.13, 29.48)

6.35 (5.99, 6.70)

Inner Tails

14.94 (14.35, 15.6) 1.26

(1.22, 1.30)

14.7 (14.32, 15.10) 1.26

(1.22, 1.30) Outer

Tails

16.87 (16.70, 16.99)

14.84 (14.44, 15.29)

Outer HG/Core

28.35 (27.66, 28.98)

53.59 (51.42, 55.97)

23.06 (22.36, 23.68)

51.77 (49.61, 54.06)

HG: Headgroup


112

Figure 4.13: Neutron reflectivity profile (A and model data fits with their scattering length density profiles (B) for an assymetrically deposited DPPC (inner leaflet) and Ra-LPS (outer leaflet) model membrane at 38 °C. The sample was measured in three isotopic contrasts (100% D2O, 100% H2O and 38% D2O/SMW). The model membrane was fitted into a five-layered mathematical model: Silicon oxide (SiO2), DPPC headgroup (Inner HG), DPPC tails (Inner tails), Ra LPS tails (Outer tails) and Ra LPS headgroups (Core).

Table 4.7: Volume fraction of deuterated DPPC tails, hydrogenous LPS tails and water within the bilayers of Ra-LPS/d-DPPC 38 °C before and after the challenge by the fluidosomes.

Layer Asymmetric bilayer at 38 °C

Before challenge After challenge 𝝋𝑫𝑷𝑷𝑪 𝝋𝑹𝒂𝑳𝑷𝑺 𝝋𝒘𝒂𝒕𝒆𝒓 𝝋𝑫𝑷𝑷𝑪 𝝋𝑹𝒂𝑳𝑷𝑺 𝝋𝒘𝒂𝒕𝒆𝒓

SiO2 n/a n/a 0.16

(0.15, 0.17) n/a n/a

0.16 (0.16, 0.17)

Inner Headgroup 0.78

(0.77, 0.79) 0.21

(0.21, 0.20) 0.013

(0.012, 0.013)

0.65 (0.64, 0.67)

0.35 (0.34, 0.31)

0.0003 (0.0003, 0.0003)

Inner Tails

Outer Tails 0.14

(0.13, 0.16) 0.85

(0.85, 0.83) 0.21

(0.20, 0.22) 0.79

(0.77, 0.78) Outer Headgroup (Core)

Next, the reflectivity profile of the model membrane after interaction with fluidosomes was

studied. The damaging effect of fluidosomes on the Rc J5 LPS, which had been seen in the

air/liquid interface experiments (Figure 4.10C) was further investigated by measuring changes in

the structure of the fully characterised d-DPPC:h-Ra-LPS bilayer after being challenged with 0.1

mg/mL empty fluidosomes for 1 h. The reflectivity profile of the system was measured after flushing out excess fluidosomes, composed of hydrogenated lipids (DPPC:DMPG). Most of the

structural changes occurring following the addition of fluidosomes involved the tail and outer

headgroup regions. The exposure of the bilayer to fluidosomes resulted in the addition of a fringe


113

in the reflectivity profile, reflecting the presence of two lipid bilayer, one being the model OM and

the second being described as floating bilayer (7th layer) in the SLD profile (Figure 4.14B).

Figure 4.14: Neutron reflectivity profile (A) and model data fits with their scattering length density profiles (B) for an assymetrically deposited DPPC (inner leaflet) and Ra-LPS (outer leaflet) model membrane after being challenged with 0.1 mg/mL fluidosomes (DPPC/DMPG, 18:1) at 38 °C. The sample was measured in three isotopic contrasts (100% D2O, 100% H2O and 38% D2O/SMW). The model membrane was fitted into a seven-layered mathematical model: Silicon oxide (SiO2), DPPC headgroup (Inner HG), DPPC tails (Inner tails), Ra LPS tails (Outer tails) and Ra LPS headgroups (Core), Bridge and Floating bilayer.

Mixing of lipids within the bilayer region occurred, as derived from the lipid composition in the

model membrane, with the inner leaflet consisting of 65% d-DPPC and 34% h-Ra LPS, while the

outer leaflet of 21% d-DPPC and 78% LPS (Table 4.7). It can be observed that the contribution

of hydrogenous phospholipids in the inner leaflet layers, which might be either h-Ra LPS or h-

phospholipids from fluidosomes, had increased from 21% to 34% after adding the fluidosomes. In addition, the fitting parameters in Table 4.8 showed an increase of the model OM roughness

from 6.3 Å to 7.1 Å, before and after the challenge, respectively, which illustrates structural

changes of the membrane. A thinner inner headgroup region (5.7 Å) with higher hydration levels

(48.44 %) and a thicker outer headgroup region (21.0 Å) with lower hydration (35.95 %) were

also observed, compared to the unchallenged membrane (Table 4.6) under the same conditions.

Of the two additional layers formed, the one adjacent to the membrane is a solvent layer (99.94%

water content) with an SLD of 1.78×10-7 (1.7224×10-7, 1.8266×10-7) Å-2, thickness of 34.6 and

roughness of 3.4 Å. The outermost layer has high water content (75.80%) with an SLD of 2.6×10-

8 (2.4608×10-8, 2.7494×10-8) Å-2. If liposomes were intact on the membrane surface, their tail

SLD would have been similar to the reported SLD of h- or d-DPPC tails (-0.4 or 7.4 ×10-6 Å-2

respectively) as published by Clifton and coworkers (226). The thickness of the additional layers

was found to be 35 and 46 Å for the 6th and 7th layer, respectively.


114

This profile may reflect fusion of liposomes in the membrane because of the close to zero SLD

of the bridge layer and the increase of hydrogenous lipid tails within the bilayer (Figure 4.15,

Table 4.7).

Table 4.8: Structural parameters, namely, thickness, hydration and roughness, of the challenged Ra-LPS/DPPC membrane at 38 °C as obtained from fits to the neutron reflectivity data shown in Figure 4.14.

Layer

Asymmetric bilayer at 38 °C

Before challenge After challenge

Thickness (Å)

% water

Roughness (Å)

Thickness (Å)

% water

Roughness (Å)

SiO2 20.13

(19.58, 20.72)

16.27 (15.56, 16.90)

4.61 (4.39, 4.86)

20.13 (19.58, 20.72)

16.27 (15.57, 16.89)

4.61 (4.39, 4.86)

Inner HG 7.41

(7.03, 7.80)

28.27 (27.13, 29.48)

6.35 (5.99, 6.70)

5.75 (5.51, 5.95)

48.44 (47.04, 49.65)

7.15 (6.84, 7.54)

Inner Tails 14.7

(14.32, 15.10) 1.26

(1.22, 1.30)

15.28 (14.81, 15.71) 0.03

(0.03, 0.03)

Outer Tails 14.84

(14.44, 15.29)

16.76 (16.50, 16.97)

Outer HG/Core 23.06

(22.36, 23.68)

51.77 (49.61, 54.06)

21.05 (20.16, 22.07)

35.97 (34.49, 37.55)

Bridge - - - 34.59

(33.43, 35.74)

99.94 (99.85, 99.99)

3.36 (3.26, 3.47)

Vesicles/Bilayer - - - 45.82

(44.89, 46.882)

75.80 (75.02, 76.73)

16.62 (15.91, 17.34)


115

Figure 4.15: Representation of the possible structural changes in the model OM system after being challenged by fluidosomes.

4.5 Discussion

In this chapter, the uptake mechanism of PPA148 and its carrier were investigated in terms of

their biophysical interaction with model systems mimicking the lipidic components of the two

membranes of the Gram negative cell envelope. The main hypothesis of this project was that a

drug-in-cyclodextrin-in-liposome formulation would enhance PPA148’s efficiency against Gram

negative bacteria by fusion with the OM, releasing the drug/CD complex. Cyclodextrin increased

PPA148’s water solubility by inclusion complex formation (Chapter 2), but it was hypothesized

that it could also enhance the drug’s transport through the IM. Fluidosomes were selected as the carrier of the drug-CD complex because it is known that they can fuse into the bacterial OM and,

thereby release their payload into the periplasmic space.

The ability of PPA148 to passively diffuse through a model OM was investigated using Rc J5

LPS and Re R595 Lipid A monolayers in a Langmuir trough at the air/liquid interface (Figure

4.10A). PPA148 has been found to interact weakly with Rc J5 LPS and showed a high affinity towards the R595 Lipid A monolayer. Although the J5-LPS monolayer was in a mixed liquid

expanded/liquid condensed state at 30 mN/m (Figure 4.5C), it produced the cross-link of the

anionic molecules in the presence of Mg2+. The steric barrier produced hindered the interaction

which presented significant deviation in terms of the affinity towards the lipids (ΔΠmax) and the

speed of the interaction (Hill slope). Hydrated oligosaccharide chains in Rc J5 LPS prevent the

entrance of hydrophobic or surface-active molecules and make the hydrophobic bilayer

inaccessible as was found by Clifton and co-workers using X-ray and neutron reflectivity

techniques (225). R595 Lipid A underwent a more condensed state at 30 mN/m owing to the absence of oligosaccharide chain, leading to a higher drug affinity observed in the pressure/time

isotherm. The smaller headgroup allows the R595 Lipid A monolayer to have more closely


116

packed molecules (Figure 4.5C), presenting a weaker steric hindrance to PPA148. Although

PPA148 interacted with R595 Lipid A, rifampicin did not show any interaction (Figure 4.10B),

which was unexpected based on its proposed mechanism of uptake. This can be explained by the topological polar surface area (tPSA), a parameter predicting the passive drug permeation

through the Gram-negative (tPSA of 165 Å2) and Gram-positive (tPSA 243 Å2) bacterial envelope

(12). It was estimated by ChemDraw that PPA148 has a tPSA of 141 Å2 (Table 2.1), which

indicates a less polar compound than rifampicin (tPSA of 217 Å2). Thus the novel drug is more

likely to interact with the R595 Lipid A, as found in the air/liquid monolayer experiments.

We also obtained evidence that PPA148 interacts with the IM lipid monolayers (Figure 4.6C,

Figure 4.8B). Despite some stability difficulties with the E. coli B monolayer (Figure 4.6C, Figure

4.7), it was found that PPA148 interacted with phospholipids, but it did not cause membrane

damage as happened with chlorhexidine (Figure 4.7). E. coli B lipid monolayer is a poor model

of the membrane because it contains a diverse mixture of lipids (Table 1.6), which, in live

bacteria, would form an asymmetric bilayer, with the lipids which impart negative curvature

localized on the inside (124,125). Instead, in a monolayer, these lipids are forced to adopt a planar configuration and, thus, the monolayer presents packing discontinuities without a

reproducible interaction profile (Figure 4.6C).

Using a monolayer model, composed of DPPC/DPPG (3:1), it was observed that PPA148 and

rifampicin presented a similar interaction profile (Figure 4.8). The findings for rifampicin are in

agreement with its mechanism of transport: that it partitions in DPPC monolayers by establishing ionic bonds with the head groups of DPPC in an acidic environment of pH=5 (255). Since the

kinetic profile of PPA148 was similar to that of rifampicin, it is suggested that PPA148 is likely to

enter the cell via diffusion through the membrane. However, PPA148 cannot be ionized (Figure

2.2) and, thus, the possibility of this type of interaction occurring with phospholipids was

discounted. It is more likely that PPA148 is partitioning into the membrane because of its

hydrophobicity.

As PPA148’s solubilizers, HPβCD and RAMEB were examined separately for their interaction

with both model OM and IM monolayers (Figure 4.9; Figure 4.11). It was found that HPβCD did

not affect the OM monolayer, while RAMEB presented a weak interaction with it, possibly

resulting from the different substituted groups (Figure 4.11). RAMEB is a methyl substituted βCD

derivative with high hydrophobicity which makes it more likely to partition into LPS. It has been

reported that complexes of drug-β-cyclodextrin and its derivatives, such as RAMEB, improve

antibiotic potency but the mechanism has not been clarified yet. It is hypothesized that β-CD may drive internalization of the β-CD-antibiotic complex via (i) the OM protein CymA (cyclodextrin

metabolism A), (ii) enhanced adhesion to the bacterial surface with potential local release of the

antibiotic, and (iii) destabilization of the bacterial envelope (215,216,246). The transport through

the CymA channels is a controversial issue seeing that some scientists have found that α-CD

and β-CD bind and can transported into the cell via those porins but others have found that only


117

the permeation of α-CD can be mediated by CymA (247) due to the channels’ cut-off size of

950Da. In the current work, porin channels have not been investigated as a mechanism of

uptake, so, based on our results, RAMEB might just enhance the adhesion to the OM.

The interaction of HPβCD and RAMEB with model IM resulted in a drop of surface pressure to

negative values (Figure 4.9), suggesting a lipid solubilizing effect of both DPPC/DPPG and

POPC/POPG monolayers. It was found that upon cyclodextrin-membrane interaction, the

packing of phospholipids at the air/liquid interface and the type of βCD derivative (hydroxy-propyl-

and methyl- substituted groups) affect the efficiency of the interaction (Figure 4.9). POPC/POPG formed a liquid condensed monolayer at 30 mN/m (Figure 4.3C) which favours the drug

membrane interaction because the lipids have lateral and rotational mobility, promoting the

molecular process of binding (Figure 4.16). Both types of cyclodextrin resulted in ΔΠmax of -20

mN/m, which suggests strong interaction with POPC/POPG by removing lipids from the surface.

DPPC/DPPG monolayer produced a condensed film at the air/liquid interface, with which HPβCD

interacted weakly. The tight packing of lipids and the saturated hydrocarbon chains hindered the

interaction. RAMEB shows a high affinity towards the DPPC/DPPG monolayer, which was expected because RAMEB has been reported to induce phospholipid exchange more efficiently

than HPβCD (181). In addition, it has also been reported that RAMEB forms soluble complexes

with DPPC (210), which might explain the negative surface pressure of the isotherm, but it was

not examined further in the current work.

Figure 4.16: Schematic representation of the movement of phospholipids within a membrane: transpose diffusion (1), rotation (2), swing (3), flexion (4) and transverse diffusion or “flip-flop” movement (248).

Fluidosomes (DPPC/DPMG, 18:1) were used as a carrier to overcome the barrier of the OM

because monolayer interaction studies showed that PPA148 cannot diffuse through the OM

model membranes (Figure 4.10A). It was hypothesized that fluidosomes could release their

content (drug-in cyclodextrin) by fusion into the OM membrane as found by Wang et al. for a

DPPC/DMPG (9:1) system by lipid mixing assay (249). The air/liquid monolayer interaction

experiment revealed a strong interaction between fluidosomes and Rc J5 LPS (Figure 4.10C). The interaction isotherm revealed a lipid solubilizing effect upon fluidosome injection in the

subphase, containing 1 mM MgCl2. Wang et al. reported that the aggregation and fusion of

fluidosomes with Gram negative bacterial membrane is promoted by the presence of Ca2+ or

other divalent cations by inducing the neutralization of the negatively charged Lipid A part and

dehydration of phospholipids of the membrane (249).


118

To further investigate the interaction between fluidosomes and OM, neutron reflectivity was used

on an asymmetric model Gram negative bacterial membrane consisting of Ra-EH100 LPS and

DPPC in the outer and inner leaflet, respectively. Before introducing fluidosomes into the system, the model membrane was characterized in terms of structural properties at room temperature

(25 °C) and 38 °C (Figure 4.12; Figure 4.13). At 38 °C lipids become more flexible (250,251) as

found in this work. The thickness of all the individual compartments of the model membrane

decreased as the temperature increased (Table 4.6). Membrane rigidity was modulated

thermodynamically by changes in temperature and decreased with increasing temperature.

Changes in orientation and hydration of lipids upon increasing temperature caused thinning of

lipid bilayers (250) because the hydrocarbon lipid chains became more flexible (Figure 4.17). The roughness of the membrane increased with increasing temperature which indicated a liquid

crystalline phase of the lipids, allowing a less rigid structure of the membrane (Table 4.6). In a

liquid crystalline membrane, lipids can simply transpose with neighboring molecules (transpose

diffusion), rotate quickly around their axis, swing from side to side, following contraction

movement or, less frequent, undergo flip-flop movement (transverse diffusion) (Figure 4.16;

Figure 4.17) (248). The lipid movement creates a more favorable environment for interaction with

the membrane.

Figure 4.17: Schematic representation of the effect of temperature change on membrane structure and behavior of lipid bilayers adapted from Los and Murata (250). Low temperatures cause “rigidification” of membranes, whereas high temperatures cause “fluidization” of membranes.

The effect of fluidosomes on the assymetric model OM was observed at 38 °C (Figure 4.14;

Figure 4.15). The NR profile of the membrane after addition of fluidosomes revealed a lipid mixing

profile with an additional bilayer associated with the deposited membrane. The roughness of the

model OM increased compared to the membrane before being challenged by the fluidosomes,

and the volume fractions of deuterated and hydrogenous tails of the inner and outer leaflets

changed, revealing a mixing of the lipids (Table 4.6; Table 4.7; Table 4.8). More specifically, the volume fraction of d62DPPC increased in the outer leaflet, while hydrogenous lipids increased in


119

the inner leaflet of the membrane. Based on the fitted parameters, the SLD of the additional layer

adjacent to the membrane was close to zero, indicating that it is composed of both deuterated

and hydrogenous material to adjust the SLD at 0.18×10-7 Å-2. The outermost additional layer presented a similar thickness (46 Å) to a DPPC bilayer (50 Å) in the presence of CaCl2 (206).

This strongly suggests that fluidosomes fused into the OM membrane by creating a Ca2+ bridge

layer between the outer leaflets of the membrane and liposome at the contact site (Figure 4.15).

Ca2+ is able to keep fluidosomes attached to the model membrane by interacting with both

liposomes and LPS accompanied by local dehydration, lowering the free energy and creating

local packing defects on the outer leaflet of both systems (Figure 4.15). Moreover, fluidosomes

tend to destabilize thermodynamically at 38 °C which also aids their fusion with the bacterial OM (252). Vesicle fusion with bacterial membranes, promoted by divalent cations, requires fluidity

which is achieved at a temperature higher than the phase transition temperature of fluidosomes

(approximately 35°C ) (249). The thinning of the model OM and increased flexibility of lipids at

38 °C (Table 4.6) facilitate the fusion of fluidosomes with the model Gram negative bacterial

membrane.

After considering both the air/liquid interface interaction and the NR experiment, it is suggested

that both fluidosomes and model OM underwent conformational changes, leading to their

collapse and fusion at the point of contact (Figure 4.15).

4.6 Conclusion

In the current study, the uptake mechanism of PPA148 and its formulation was studied using

biophysical interfacial techniques. The monolayer studies suggest that PPA148 slowly diffuses

through the lipidic components of the PC/PG model IM while it presents weak interaction with the

model OM consisting of DPPC and Ra-EH100 LPS as the inner and outer leaflet, respectively.

When compared to the pressure-time isotherm profile of chlorhexidine, rifampicin and gentamicin, whose mechanism of uptake is known, PPA148 showed a similar profile with that of

rifampicin, indicating that PPA148 passively diffuses through the IM. The interaction of PPA148

with the LPS monolayer, characterized by slow kinetics, suggests that the main mechanism of

PPA148’s transport through the OM is dependent on non-specific or self-promoted diffusion or

even active transport. In this chapter, it was found that cyclodextrins might act as permeation

enhancers through the IM, in addition to increasing the drug’s water solubility. The structural

changes occurring in the model membrane upon being exposed to fluidosomes were examined by neutron reflectivity and suggest that fluidosomes enter into the LPS leaflet of the OM through

a fusion mechanism.

General Conclusion

120

Chapter 5 General Conclusion

General Conclusion

121

Experience has demonstrated that bacteria are quick to develop resistance to new antibiotics.

This predictable resistance is essentially related to the mechanism of action of traditional

antibiotics with five main specific targets in the cell: cell wall synthesis, DNA-gyrase, DNA-directed-RNA-polymerase, protein synthesis and enzymes (253), and the speed at which

mutation may facilitate target alteration. The development of new classes of antibiotics with

specificity towards bacteria, but which act non-specifically against novel targets, needs to be

pursued as a possible method to delay the onset of resistance. At King’s College London, an antimicrobial compound, PPA148, was synthesized, which is active against Gram-negative bacteria with a novel antimicrobial mechanism of action. PPA148’s

bactericidal activity is attributed to its binding onto the minor-groove of bacterial DNA. It showed

promising activity against the Gram negative members of the ESKAPE organisms, and little

toxicity to human cells (107,108). Nevertheless, in some clinical strains, difficulties in drug

uptake, due to the bacterial semi-permeable OM and/or the efflux mechanisms, was observed.

It is known that intracellular drug accumulation is a complex process including drug uptake into

the cell, retention and distribution in the cell, and efflux from the cell. At any given time, the accumulation of a drug in cells is dependent upon the different rates of drug uptake and efflux

(254). Overexpression of efflux mechanisms leads to a decrease in drug accumulation resulting

in drug-resistant cells. Based on our knowledge of the physicochemical properties of PPA148, reducing particle

aggregation and enhancing water solubility was the first step towards developing a formulation to enhance its antibacterial efficacy against Gram negative bacteria. The solubility increased 6-

fold by incorporating PPA148 into cyclodectrin’s cavity (Table 2.3; Table 2.4). Fluorescence

spectroscopy was used to determine the binding constant of a 1:1 complex with HPβCD and

RAMEB, which was found to be 63 ± 20 M-1 and 102 ± 26 M-1, respectively (Table 2.4). Evidence

of the formation of inclusion complex include NMR spectroscopy, which presented a shift of the

DIMEB proton at position 5 protruding towards the CD cavity (Figure 2.19). The results revealed

the formation of a 1:2 PPA148/DIMEB complex.

The next step was to incorporate the 1:1 PPA148/RAMEB complex into fluidosomes

(DPPC/DMPG, 18:1) and test the efficacy of the drug in its pure and formulated form. The

entrapment of the water-soluble inclusion complex into fluidosomes led to accommodation of the

insoluble PPA148 in the aqueous phase of neutral vesicles with a size of 129 ±10 nm and high

encapsulation efficiency (67 ± 11%). Loaded fluidosomes presented a more monodisperse

system than the empty fluidosomes (Figure 3.4; Figure 3.5), indicating that the presence of the complex created structures with increased integrity. Generally, in liposomes, cyclodextrin

complexation competes with liposomal membrane binding, which in our case led to a more intact

system because the affinity of the drug towards CD is higher than that of lipids. The LT

experiment showed weak interaction of RAMEB with a DPPC/DPPG monolayer (Figure 4.9),

which describes interrelation with the lipid headgroups without causing any damage to the

General Conclusion

122

membrane. PPA148 might be inserted into the CD cavity from one side while the other side of

RAMEB can be attached with the headgroups of the inner leaflet of the liposomes, thus forming

more stable vesicles (Figure 3.10).

The efficacy of PPA148 against the E. coli DH5α (PPA148-in-RAMEB-in-fluidosomes) increased

when it was complexed with RAMEB. Pure PPA148 created an inhibition zone of 1.6 ± 0.2 cm,

while the complex generated a 2.4 ± 0.1 cm inhibition zone (Figure 3.7). The difference was

statically significant (Figure 3.8), thus showing that the efficacy of PPA148 was enhanced.

Therefore, RAMEB has a dual role in (1) increasing the solubility of the PPA148 and (2) transporting drug molecules across membranes (intrinsic antimicrobial activity). The efficacy was

also enhanced when the complex was incorporated into fluidosomes with an inhibition zone of

2.7 ± 0.2 cm which was statistically significantly different from the pure drug; but the difference

between the inhibition zone of the complex and the final formulation (Figure 3.8) was not

statistically significant, which might be a result of the small number of experimental observations

To further elucidate the efficiency of the final formulation, Monte Carlo simulation was performed

by building a probability distribution and creating a substituted set of data (100 observations) based on the experimental observations. Predicted data were analyzed, and it was estimated

that the difference between the efficacy of the final formulation and the complex would be

statistically significant (Figure 3.9). This preliminary microbiological test presented in this thesis

has some statistical limitations due to the small number of observations, nevertheless it shows

that the final formulation increased the antimicrobial activity of the novel compound.

Following the microbiological assay, a biophysical approach was pursued to investigate the

mechanism of uptake of PPA148 and its carriers. Knowing the physicochemical properties of

PPA148, passive diffusion through lipid monolayers was hypothesized as a possible mechanism

of uptake. The LT experiments at the air/liquid interface revealed that PPA148 slowly diffuses

through both the IM (DPPC/DPPG) (Figure 4.8B) and OM (Rc J5 LPS and R595 Lipid A) (Figure

4.10A) model monolayers. The use of Lipid A and LPS revealed the steric barrier created by the

oligosaccharide chains which hinders the permeation of PPA148 through the OM. These results,

in combination with the inhibition caused in live bacteria, reveal that the main mechanism of PPA148’s transport is dependent on non-specific or self-promoted diffusion or even active

transport.

HPβCD and RAMEB presented a strong interaction with POPC/POPG monolayers but a weak

interaction with those formed from DPPC/DPPG (Figure 4.9). Neither was cyclodextrin able to

interact strongly with the OM model membrane (Figure 4.11), yet it enhanced the antimicrobial efficacy of the drug when tested against E. coli DH5α live bacteria. These results reveal that

RAMEB is able to potentiate the transport of PPA148 by improving the adhesion to the bacterial

surface with potential local release of the antibiotic through the IM, without causing any damage

to the membrane.

General Conclusion

123

It is suggested that fluidosomes act as drug carriers, releasing their content via fusion with the

bacterial OM. Langmuir trough experiments at the air/liquid interface presented a strong

interaction between fluidosomes and Rc J5 LPS (Figure 4.10C), revealing the removal of lipids from the surface. This behaviour was investigated in depth using neutron reflectivity, which

provides a better insight into the mechanism of action of fluidosomes focusing at the molecular

level on the localization of vesicles interacting with membrane models. Fluidosomes challenged

an assymetric Gram negative bacterial OM, consisting of Ra EH100 LPS and d62-DPPC as the

outer and inner leaflet, respectively (Figure 4.14). The results suggested the mixing of lipids

induced by fluidosomes because the volume fraction of d62-DPPC increased in the outer leaflet,

while hydrogenous lipids increased in the inner leaflet of the membrane (Table 4.7). The NR profile presented two additional layers attached to the model membrane, the one describing a

water layer connecting the membrane with fluidosomes and the other being the bilayer of the

fluidosomes (Figure 4.14; Figure 4.15).

Overall, the data gathered in this thesis has shown that drug-in-RAMEB-in-fluidosomes is a

promising formulation for antimicrobial applications and should be tested in a range of Gram negative bacterial membrane. This work has demonstrated that fluidosomes fuse into the model

bacterial OM, facilitating the release of its core content (Figure 4.15) and derivatives of βCD

(HPβCD and RAMEB) enable possibly the transport through the IM. This work is also the first

characterization of the novel antimicrobial compound, PPA148, in terms of solubility,

environmental stability (temperature and pH) and spectroscopic profile.

5.1.1 Future Work

This thesis was an attempt to characterize a novel antimicrobial compound, and enhance its

water solubility and efficacy against Gram negative bacteria. The hypotheses of this project were

driven by PPA148’s predicted physicochemical properties and its MIC against a range of

bacterial strains. The MIC of pure PPA148 presented a drop in the presence of PAβN, which

may have been either due to its efflux pump inhibitory action or its bacterial membrane

permeabilizing behavior. Therefore, future work should include the assessment of the integrity of

bacterial cell membranes using the fluorescent probe, 8-anilino-1-naphthylenesulfonic acid (ANS). ANS is a neutrally charged, hydrophobic probe that fluoresces weakly in aqueous

environments, but exhibits enhanced fluorescence in nonpolar/hydrophobic environment (255).

This will give an insight on the effect of PAβN on PPA148’s antibacterial activity which in turn will

provide invaluable information about the drug’s mechanism of action. The interaction studies

would provide structural insights on the complexes and the spherical morphology of the loaded

liposomes (110). Future work should also include a drug release profile of PPA148, or any other

antimicrobial compound encapsulated in cyclodextrin/liposome formulation to determine the

degree of drug leakage over time. The dialysis method is applied widely and drug analysis with UV/Vis spectroscopy will identify the released drug concentration. Stability of the drug in the

General Conclusion

124

formulation over time is another important aspect that needs to be defined using mass

spectroscopy to detect any possible decomposed products.

In this study, the fusion of fluidosomes with the lipidic component of a model assymetric Gram

negative bacterial membrane was observed using neutron reflectivity. Development of advanced

models of Gram negative membranes containing proteins are in progress (256). Using these

advanced model membranes to investigate the fusion will give an insight on whether proteins

affect this mechanism of action. For a more realistic approach, in vitro experiments on live

bacteria should be applied using spectroscopic techniques to localize the formulation within the bacterial cell. Even though the lipid composition in eukaryotic and prokaryotic cells is different

(223,224), the toxicity profile of fluidosomes should also be investigated with human epithelial

cells to ascertain whether there are any possible side effects for host cells.

Disk diffusion assay was used to assess the susceptibility of bacteria to PPA148-in-RAMEB-in-

liposomes. The microbiological assay was focused only on one bacterial strain (E coli DH5α) yet, to obtain more conclusive data regarding the efficacy of the final formulation on Gram negative

bacteria, it must be tested across a wide range of antibiotic-resistant bacteria, such as P.

aeruginosa. In this test, which is qualitative, the diameter of the inhibition zone is related to the

susceptibility of the isolate and to the diffusion rate of the drug through the agar medium. The

drawback of this method is that the category of susceptibility is derived from the test rather than

from the minimum inhibitory concentration (MIC). That being the case, the MIC of the final

formulation needs to be determined in the different types of bacteria used. Formulation with other antibiotics, whose target is in the cytoplasm, i.e. rifampicin, and have become ineffective due to

the rise of antimicrobial resistance, should also be tested to elucidate the effectiveness of the

drug-in-cyclodextrin-in-liposomes formulation. Collecting all the aforementioned data will help to

elucidate the antibacterial activity of a drug-in-cyclodextrin-in-fluidosomes.

References

125

References 1. Livermore DM, Blaser M, Carrs O, Cassell G, Fishman N, Guidos R, et al. Discovery

research: the scientific challenge of finding new antibiotics. J Antimicrob Chemother. 2011 Sep;66(9):1941–1944.

2. Ventola CL. The antibiotic resistance crisis: part 1: causes and threats. P T Peer-Rev J Formul Manag. 2015 Apr;40(4):277–83.

3. Gould IM, Bal AM. New antibiotic agents in the pipeline and how they can help overcome microbial resistance. Virulence. 2013 Feb;4(2):185–91.

4. Wright GD. Something old, something new: revisiting natural products in antibiotic drug discovery. Can J Microbiol. 2014 Mar;60(3):147–154.

5. Rossolini GM, Arena F, Pecile P, Pollini S. Update on the antibiotic resistance crisis. Curr Opin Pharmacol. 2014 Oct;18:56–60.

6. Piddock LJ. The crisis of no new antibiotics—what is the way forward? Lancet Infect Dis. 2012 Mar;12(3):249–253.

7. Yocum RR, Rasmussen JR, Strominger JL. The mechanism of action of penicillin. Penicillin acylates the active site of Bacillus stearothermophilus D-alanine carboxypeptidase. J Biol Chem. 1980 May;255(9):3977–86.

8. Wehrli W. Rifampin: mechanisms of action and resistance. Rev Infect Dis. 1983;5 Suppl 3:S407–11.

9. Levy SB, Marshall B. Antibacterial resistance worldwide: causes, challenges and responses. Nat Med. 2004 Dec;10(12s):S122–S129.

10. Clatworthy AE, Pierson E, Hung DT. Targeting virulence: a new paradigm for antimicrobial therapy. Nat Chem Biol. 2007 Sep;3(9):541–548.

11. van Saene R, Fairclough S, Petros A. Broad- and narrow-spectrum antibiotics: a different approach. Clin Microbiol Infect Off Publ Eur Soc Clin Microbiol Infect Dis. 1998 Jan;4(1):56–57.

12. O’Shea R, Moser HE. Physicochemical Properties of Antibacterial Compounds: Implications for Drug Discovery. J Med Chem. 2008 May;51(10):2871–2878.

13. Macielag MJ. Chemical Properties of Antimicrobials and Their Uniqueness. In: Dougherty T, Pucci M, editors. Antibiotic Discovery and Development. Boston, MA: Springer US; 2012. p. 793–820.

14. Leeson PD, Davis AM. Time-Related Differences in the Physical Property Profiles of Oral Drugs. 2004;47(25):6338–6348.

15. ChemBioDraw Ultra 14.0 Suite - Adept Scientific [Internet]. [cited 2018 Oct 6]. Available from: http://www.adeptscience.co.uk/products/lab/chembiodraw/chembiodraw-ultra-suite.html

16. Fernandes J, Gattass CR. Topological Polar Surface Area Defines Substrate Transport by Multidrug Resistance Associated Protein 1 (MRP1/ABCC1). J Med Chem. 2009 Feb;52(4):1214–1218.

References

126

17. Hou TJ, Zhang W, Xia K, Qiao XB, Xu XJ. ADME Evaluation in Drug Discovery. 5. Correlation of Caco-2 Permeation with Simple Molecular Properties. J Chem Inf Comput Sci. 2004 Sep;44(5):1585–1600.

18. Silhavy TJ, Kahne D, Walker S. The bacterial cell envelope. Cold Spring Harb Perspect Biol. 2010 May;2(5):a000414.

19. Galanos C, Freudenberg MA. Bacterial endotoxins: biological properties and mechanisms of action. Mediators Inflamm. 1993;2(7):S11–16.

20. Homma JY, Matsuura M, Kanegasaki S, Kawakubo Y, Kojima Y, Shibukawa N, et al. Structural requirements of lipid A responsible for the functions: a study with chemically synthesized lipid A and its analogues. J Biochem (Tokyo). 1985 Aug;98(2):395–406.

21. Kotani S, Takada H, Tsujimoto M, Ogawa T, Takahashi I, Ikeda T, et al. Synthetic lipid A with endotoxic and related biological activities comparable to those of a natural lipid A from an Escherichia coli re-mutant. Infect Immun. 1985 Jul;49(1):225–237.

22. Merck KGaA. Lipopolysaccharides - Structure, Function and Application | Sigma-Aldrich [Internet]. 2018 [cited 2018 Oct 20]. Available from: https://www.sigmaaldrich.com/technical-documents/articles/biology/glycobiology/lipopolysaccharides.html

23. Rivera M, McGroarty EJ. Analysis of a common-antigen lipopolysaccharide from Pseudomonas aeruginosa. J Bacteriol. 1989 Apr;171(4):2244–2248.

24. Steimle A, Autenrieth IB, Frick J-S. Structure and function: Lipid A modifications in commensals and pathogens. Int J Med Microbiol. 2016 Aug;306(5):290–301.

25. Wydro P, Flasiński M, Broniatowski M. Molecular organization of bacterial membrane lipids in mixed systems—A comprehensive monolayer study combined with Grazing Incidence X-ray Diffraction and Brewster Angle Microscopy experiments. Biochim Biophys Acta BBA - Biomembr. 2012 Jul;1818(7):1745–1754.

26. Piddock LJV. Clinically Relevant Chromosomally Encoded Multidrug Resistance Efflux Pumps in Bacteria. Clin Microbiol Rev. 2006 Apr;19(2):382–402.

27. Livermore DM. Antibiotic uptake and transport by bacteria. Scand J Infect Dis Suppl. 1990;74:15–22.

28. Tipper DJ. Mode of action of beta-lactam antibiotics. Pharmacol Ther. 1985;27(1):1–35.

29. Fernandez L, Hancock REW. Adaptive and Mutational Resistance: Role of Porins and Efflux Pumps in Drug Resistance. Clin Microbiol Rev. 2012 Oct;25(4):661–681.

30. Vila J, Martí S, Sánchez-Céspedes J. Porins, efflux pumps and multidrug resistance in Acinetobacter baumannii. J Antimicrob Chemother. 2007 Jun;59(6):1210–1215.

31. Pajatsch M, Andersen C, Mathes A, Böck A, Benz R, Engelhardt H. Properties of a cyclodextrin-specific, unusual porin from Klebsiella oxytoca. J Biol Chem. 1999 Aug;274(35):25159–66.

32. Taber HW, Mueller JP, Miller PF, Arrow AS. Bacterial uptake of aminoglycoside antibiotics. Microbiol Rev. 1987 Dec;51(4):439–457.

33. Nikaido H. Molecular basis of bacterial outer membrane permeability revisited. Microbiol Mol Biol Rev MMBR. 2003 Dec;67(4):593–656.

References

127

34. Bolla J-M, Alibert-Franco S, Handzlik J, Chevalier J, Mahamoud A, Boyer G, et al. Strategies for bypassing the membrane barrier in multidrug resistant Gram-negative bacteria. FEBS Lett. 2011 Jun;585(11):1682–1690.

35. Amaral L, Martins A, Spengler G, Molnar J. Efflux pumps of Gram-negative bacteria: what they do, how they do it, with what and how to deal with them. Front Pharmacol. 2014;4:168.

36. Lewis K. New approaches to antimicrobial discovery. Biochem Pharmacol. 2017 Jun;134:87–98.

37. Kennedy DA, Read AF. Why does drug resistance readily evolve but vaccine resistance does not? Proc Biol Sci. 2017 Mar;284(1851):20162562.

38. Barber M, Whitehead JEM. Bacteriophage types in penicillin-resistant staphylococcal infection. Br Med J. 1949 Sep;2(4627):565–9.

39. Zaman SB, Hussain MA, Nye R, Mehta V, Mamun KT, Hossain N. A Review on Antibiotic Resistance: Alarm Bells are Ringing. Cureus. 2017 Jun;9(6):e1403.

40. Högberg LD, Heddini A, Cars O. The global need for effective antibiotics: challenges and recent advances. Trends Pharmacol Sci. 2010 Nov;31(11):509–515.

41. Dahal RH, Chaudhary DK. Microbial Infections and Antimicrobial Resistance in Nepal: Current Trends and Recommendations. Open Microbiol J. 2018;12:230–242.

42. Antibiotic / Antimicrobial Resistance | CDC [Internet]. [cited 2018 Oct 6]. Available from: https://www.cdc.gov/drugresistance/index.html

43. European Antimicrobial Resistance Surveillance Network (EARS-Net) [Internet]. [cited 2018 Oct 6]. Available from: https://ecdc.europa.eu/en/about-us/partnerships-and-networks/disease-and-laboratory-networks/ears-net

44. WHO | Antimicrobial resistance. WHO [Internet]. 2018; Available from: http://www.who.int/antimicrobial-resistance/en/

45. Delcour AH. Outer membrane permeability and antibiotic resistance. Biochim Biophys Acta. 2009;1794(5):808–16.

46. Pagès J-M, Masi M, Barbe J. Inhibitors of efflux pumps in Gram-negative bacteria. Trends Mol Med. 2005 Aug;11(8):382–9.

47. Webber MA, Piddock LJV. The importance of efflux pumps in bacterial antibiotic resistance. J Antimicrob Chemother. 2003 Jan;51(1):9–11.

48. Rice LB. Challenges in Identifying New Antimicrobial Agents Effective for Treating Infections with Acinetobacter baumannii and Pseudomonas aeruginosa. Clin Infect Dis. 2006;43:100–105.

49. Meletis G. Carbapenem resistance: overview of the problem and future perspectives. Ther Adv Infect Dis. 2016 Feb;3(1):15–21.

50. Okdah L, Le Page S, Olaitan AO, Dubourg G, Hadjadj L, Rolain J-M. New therapy from old drugs: synergistic bactericidal activity of sulfadiazine with colistin against colistin-resistant bacteria, including plasmid-mediated colistin-resistant mcr-1 isolates. Int J Antimicrob Agents. 2018 May;51(5):775–783.

51. Watkins RR, Bonomo RA. β-Lactam Antibiotics. Infect Dis. 2017 Jan;1203–1216.e2.

References

128

52. O’neil J. Securing New Drugs For Future Generations: The Pipeline Of Antibiotics [Internet]. 2015. Available from: https://amr-review.org/sites/default/files/SECURING NEW DRUGS FOR FUTURE GENERATIONS FINAL WEB_0.pdf

53. Tacconelli E, Carrara E, Savoldi A, Kattula D, Burkert F. Global Priority List Of Antibiotic-Resistant Bacteria To Guide Research, Discovery, And Development Of New Antibiotics [Internet]. 2017. Available from: http://www.cdc.gov/drugresistance/threat-report-2013/

54. Bassetti M, Righi E. New antibiotics and antimicrobial combination therapy for the treatment of gram-negative bacterial infections. Curr Opin Crit Care. 2015 Oct;21(5):402–411.

55. Tängdén T. Combination antibiotic therapy for multidrug-resistant Gram-negative bacteria. Ups J Med Sci. 2014 May;119(2):149–53.

56. Worthington RJ, Melander C. Combination approaches to combat multidrug-resistant bacteria. Trends Biotechnol. 2013 Mar;31(3):177–84.

57. WHO publishes list of bacteria for which new antibiotics are urgently needed [Internet]. [cited 2018 Oct 6]. Available from: http://www.who.int/news-room/detail/27-02-2017-who-publishes-list-of-bacteria-for-which-new-antibiotics-are-urgently-needed

58. Tamma PD, Cosgrove SE, Maragakis LL. Combination Therapy for Treatment of Infections with Gram-Negative Bacteria. Clin Microbiol Rev. 2012 Jul;25(3):450–470.

59. Tascini C, Tagliaferri E, Giani T, Leonildi A, Flammini S, Casini B, et al. Synergistic activity of colistin plus rifampin against colistin-resistant KPC-producing Klebsiella pneumoniae. Antimicrob Agents Chemother. 2013 Aug;57(8):3990–3.

60. Melander RJ, Melander C. The Challenge of Overcoming Antibiotic Resistance: An Adjuvant Approach? ACS Infect Dis. 2017;3(8):559–563.

61. Rolinson GN. The history and background of Augmentin. South Afr Med J Suid-Afr Tydskr Vir Geneeskd. 1982 Jul;62(5 Spec No):3A–4A.

62. Coates ARM, Hu Y. Novel approaches to developing new antibiotics for bacterial infections. Br J Pharmacol. 2007 Dec;152(8):1147–54.

63. Pagès J-M, Amaral L. Mechanisms of drug efflux and strategies to combat them: Challenging the efflux pump of Gram-negative bacteria. Biochim Biophys Acta BBA - Proteins Proteomics. 2009 May;1794(5):826–833.

64. Torres A, Zhong N, Pachl J, Timsit J-F, Kollef M, Chen Z, et al. Ceftazidime-avibactam versus meropenem in nosocomial pneumonia, including ventilator-associated pneumonia (REPROVE): a randomised, double-blind, phase 3 non-inferiority trial. Lancet Infect Dis. 2018 Mar;18(3):285–295.

65. Castanheira M, Rhomberg PR, Flamm RK, Jones RN. Effect of the β-Lactamase Inhibitor Vaborbactam Combined with Meropenem against Serine Carbapenemase-Producing Enterobacteriaceae. Antimicrob Agents Chemother. 2016 Sep;60(9):5454–5458.

66. Zhanel GG, Chung P, Adam H, Zelenitsky S, Denisuik A, Schweizer F, et al. Ceftolozane/Tazobactam: A Novel Cephalosporin/β-Lactamase Inhibitor Combination with Activity Against Multidrug-Resistant Gram-Negative Bacilli. Drugs. 2014 Jan;74(1):31–51.

67. Lomovskaya O, Warren MS, Lee A, Galazzo J, Fronko R, Lee M, et al. Identification and Characterization of Inhibitors of Multidrug Resistance Efflux Pumps in Pseudomonas

References

129

aeruginosa: Novel Agents for Combination Therapy. Antimicrob Agents Chemother. 2001 Jan;45(1):105–116.

68. Lomovskaya O, Bostian KA. Practical applications and feasibility of efflux pump inhibitors in the clinic—A vision for applied use. Biochem Pharmacol. 2006 Mar;71(7):910–918.

69. Lamers RP, Cavallari JF, Burrows LL. The Efflux Inhibitor Phenylalanine-Arginine Beta-Naphthylamide (PAβN) Permeabilizes the Outer Membrane of Gram-Negative Bacteria. PLOS ONE. 2013 Mar 27;8(3):e60666.

70. Lewis K. Platforms for antibiotic discovery. Nat Rev Drug Discov. 2013 May;12(5):371–387.

71. Edwards DI. Nitroimidazole drugs–action and resistance mechanisms. I. Mechanisms of action. J Antimicrob Chemother. 1993 Jan;31(1):9–20.

72. Fleck LE, North EJ, Lee RE, Mulcahy LR, Casadei G, Lewis K. A Screen for and Validation of Prodrug Antimicrobials. Antimicrob Agents Chemother. 2014 Mar;58(3):1410–1419.

73. Brewster ME, Loftsson T. Cyclodextrins as pharmaceutical solubilizers. Adv Drug Deliv Rev. 2007 Jul;59(7):645–666.

74. French D. The Schardinger Dextrins. Adv Carbohydr Chem. 1957 Jan;12:189–260.

75. Loftsson T, Másson M, Brewster ME. Self-Association of Cyclodextrins and Cyclodextrin Complexes. J Pharm Sci. 2004 May;93(5):1091–1099.

76. Biwer A, Antranikian G, Heinzle E. Enzymatic production of cyclodextrins. Appl Microbiol Biotechnol. 2002 Sep;59(6):609–617.

77. Ryzhakov A, Do Thi T, Stappaerts J, Bertoletti L, Kimpe K, Sá Couto AR, et al. Self-Assembly of Cyclodextrins and Their Complexes in Aqueous Solutions. J Pharm Sci. 2016 Sep;105(9):2556–2569.

78. González-Gaitano G, Rodríguez P, Isasi JR, Fuentes M, Tardajos G, Sánchez M. The Aggregation of Cyclodextrins as Studied by Photon Correlation Spectroscopy. J Incl Phenom Macrocycl Chem. 2002;44(1/4):101–105.

79. Santosh S. Terdale, Dagade and DH, Patil* KJ. Thermodynamic Studies of Molecular Interactions in Aqueous α-Cyclodextrin Solutions: Application of McMillan−Mayer and Kirkwood−Buff Theories. 2006;110(37):18583–18593.

80. Coleman AW, Nicolis I, Keller N, Dalbiez JP. Aggregation of cyclodextrins: An explanation of the abnormal solubility of?-cyclodextrin. J Incl Phenom Mol Recognit Chem. 1992 Jun;13(2):139–143.

81. Zhang J, Ma PX. Cyclodextrin-based supramolecular systems for drug delivery: Recent progress and future perspective. Adv Drug Deliv Rev. 2013 Aug;65(9):1215–1233.

82. Sharma N, Baldi A. Exploring versatile applications of cyclodextrins: an overview. Drug Deliv. 2016 Jul;23(3):1–19.

83. Gidwani B, Vyas A. Inclusion complexes of bendamustine with β-CD, HP-β-CD and Epi-β-CD: in-vitro and in-vivo evaluation. Drug Dev Ind Pharm. 2015 Dec;41(12):1978–1988.

84. GRAS Notices [Internet]. 2016 [cited 2018 Oct 6]. Available from: https://www.accessdata.fda.gov/scripts/fdcc/index.cfm?set=GRASNotices&sort=Date_of_closure&order=DESC&showAll=true&type=basic&search=

References

130

85. Medicines Agency E. Background review for cyclodextrins used as excipients [Internet]. 2014. Available from: www.ema.europa.eu/contact

86. Annex to the European Commission guideline on “Excipients in the labelling and package leaflet of medicinal products for human use” (SANTE-2017-11668) Excipients and information for the package leaflet [Internet]. 2017. Available from: https://www.ema.europa.eu/documents/scientific-guideline/annex-european-commission-guideline-excipients-labelling-package-leaflet-medicinal-products-human_en.pdf

87. Szente, Szejtli. Highly soluble cyclodextrin derivatives: chemistry, properties, and trends in development. Adv Drug Deliv Rev. 1999 Mar;36(1):17–28.

88. Uekama K, Otagiri M. Cyclodextrins in drug carrier systems. Crit Rev Ther Drug Carrier Syst. 1987;3(1):1–40.

89. Tiwari G, Tiwari R, Rai AK. Cyclodextrins in delivery systems: Applications. J Pharm Bioallied Sci. 2010 Apr;2(2):72–79.

90. Shastri VR, Yue I, Hildgen P, Sinistera RD, Langer R. Method of increasing the efficacy Of Antibiotics By Compexing With Cyclodextrins. US2003/0078215A1., 2003.

91. Agnes M, Thanassoulas A, Stavropoulos P, Nounesis G, Miliotis G, Miriagou V, et al. Designed positively charged cyclodextrin hosts with enhanced binding of penicillins as carriers for the delivery of antibiotics: The case of oxacillin. Int J Pharm. 2017 Oct;531(2):480–491.

92. Thatiparti TR, von Recum HA. Cyclodextrin Complexation for Affinity-Based Antibiotic Delivery. Macromol Biosci. 2010 Jan;10(1):82–90.

93. Taha M, Chai F, Blanchemain N, Neut C, Goube M, Maton M, et al. Evaluation of sorption capacity of antibiotics and antibacterial properties of a cyclodextrin-polymer functionalized hydroxyapatite-coated titanium hip prosthesis. Int J Pharm. 2014 Dec;477(1–2):380–389.

94. Paczkowska M, Mizera M, Szymanowska-Powa\lowska D, Lewandowska K, B\laszczak W, Gościańska J, et al. β-Cyclodextrin complexation as an effective drug delivery system for meropenem. Eur J Pharm Biopharm. 2016 Feb;99:24–34.

95. Ferrari F, Sorrenti M, Rossi S, Catenacci L, Sandri G, Bonferoni MC, et al. Vancomycin–Triacetyl Cyclodextrin Interaction Products for Prolonged Drug Delivery. Pharm Dev Technol. 2008 Jan;13(1):65–73.

96. Deygen IM, Egorov AM, Kudryashova EV. Structure and stability of fluoroquinolone-(2-hydroxypropyl)-β-cyclodextrin complexes as perspective antituberculosis drugs. Mosc Univ Chem Bull. 2016 Jan;71(1):1–6.

97. Jones MN, Song Y-H, Kaszuba M, Reboiras MD. The Interaction of Phospholipid Liposomes with Bacteria and Their Use in the Delivery of Bactericides. J Drug Target. 1997 Jan;5(1):25–34.

98. Rukavina Z, Vanić Ž. Current Trends in Development of Liposomes for Targeting Bacterial Biofilms. Pharmaceutics. 2016 May;8(2):18.

99. Schiffelers R, Storm G, Bakker-Woudenberg I. Liposome-encapsulated aminoglycosides in pre-clinical and clinical studies. J Antimicrob Chemother. 2001 Sep;48(3):333–344.

100. Meers P, Neville M, Malinin V, Scotto AW, Sardaryan G, Kurumunda R, et al. Biofilm penetration, triggered release and in vivo activity of inhaled liposomal amikacin in chronic

References

131

Pseudomonas aeruginosa lung infections. J Antimicrob Chemother. 2008 Feb;61(4):859–868.

101. Beaulac C, Sachetelli S, Lagacé J. Aerosolization of Low Phase Transition Temperature Liposomal Tobramycin as a Dry Powder in an Animal Model of Chronic Pulmonary Infection Caused by Pseudomonas aeruginosa. J Drug Target. 1999 Jan;7(1):33–41.

102. Beaulac C, Sachetelli S, Lagace J. In-vitro bactericidal efficacy of sub-MIC concentrations of liposome- encapsulated antibiotic against gram-negative and gram-positive bacteria. J Antimicrob Chemother. 1998 Jan;41(1):35–41.

103. Beaulac C, Clément-Major S, Hawari J, Lagacé J. Eradication of mucoid Pseudomonas aeruginosa with fluid liposome-encapsulated tobramycin in an animal model of chronic pulmonary infection. Antimicrob Agents Chemother. 1996 Mar;40(3):665–669.

104. Halwani M, Yebio B, Suntres ZE, Alipour M, Azghani AO, Omri A. Co-encapsulation of gallium with gentamicin in liposomes enhances antimicrobial activity of gentamicin against Pseudomonas aeruginosa. J Antimicrob Chemother. 2008 Sep;62(6):1291–1297.

105. Manca ML, Sinico C, Maccioni AM, Diez O, Fadda AM, Manconi M. Composition influence on pulmonary delivery of rifampicin liposomes. Pharmaceutics. 2012 Nov;4(4):590–606.

106. WHO. ANTIBACTERIAL AGENTS IN CLINICAL DEVELOPMENT [Internet]. 2017. Available from: http://www.who.int/about/licensing.

107. Rahman KM, Sutton JM, Picconi P. Pbd antibacterial agents. WO 2017/098257 Al, 2017.

108. Picconi P, Miraz Rahman Anthony Coates Mark Sutton K. Design, Synthesis and Biological Evaluation of Minor Groove Binders as Antibacterial Agents Against Multi-Drug Resistant Pathogens. 2017;

109. Park DR. The microbiology of ventilator-associated pneumonia. Respir Care. 2005 Jun;50(6):742–63; discussion 763–5.

110. Zhang L, Zhang Q, Wang X, Zhang W, Lin C, Chen F, et al. Drug-in-cyclodextrin-in-liposomes: A novel drug delivery system for flurbiprofen. Int J Pharm. 2015 Aug;492(1–2):40–45.

111. McCormack B, Gregoriadis G. Entrapment of Cyclodextrin-Drug Complexes into Liposomes: Potential Advantages in Drug Delivery. J Drug Target. 1994 Jan;2(5):449–454.

112. Sebaaly C, Charcosset C, Fourmentin S, Greige-Gerges H. Potential Applications of Cyclodextrin Inclusion Complexes, Liposomes, and Drug-in-Cyclodextrin-in-Liposome in Food Industry and Packaging. Role Mater Sci Food Bioeng. 2018 Jan;187–234.

113. Azzi J, Auezova L, Danjou P-E, Fourmentin S, Greige-Gerges H. First evaluation of drug-in-cyclodextrin-in-liposomes as an encapsulating system for nerolidol. Food Chem. 2018 Jul;255:399–404.

114. Gharib R, Greige-Gerges H, Fourmentin S, Charcosset C, Auezova L. Liposomes incorporating cyclodextrin–drug inclusion complexes: Current state of knowledge. Carbohydr Polym. 2015 Sep;129:175–186.

115. Maestrelli F, González-Rodríguez ML, Rabasco AM, Mura P. Effect of preparation technique on the properties of liposomes encapsulating ketoprofen–cyclodextrin complexes aimed for transdermal delivery. Int J Pharm. 2006 Apr;312(1–2):53–60.

References

132

116. Fatouros DG, Hatzidimitriou K, Antimisiaris SG. Liposomes encapsulating prednisolone and prednisolone-cyclodextrin complexes: comparison of membrane integrity and drug release. Eur J Pharm Sci Off J Eur Fed Pharm Sci. 2001 Jun;13(3):287–96.

117. Bragagni M, Maestrelli F, Mennini N, Ghelardini C, Mura P. Liposomal formulations of prilocaine: effect of complexation with hydroxypropyl-\s s-cyclodextrin on drug anesthetic efficacy. J Liposome Res. 2010 Dec;20(4):315–322.

118. Hagiwara Y, Arima H, Hirayama F, Uekama K. Prolonged Retention of Doxorubicin in Tumor Cells by Encapsulation of γ-Cyclodextrin Complex in Pegylated Liposomes. J Incl Phenom Macrocycl Chem. 2006 Oct;56(1–2):65–68.

119. Dhule SS, Penfornis P, Frazier T, Walker R, Feldman J, Tan G, et al. Curcumin-loaded γ-cyclodextrin liposomal nanoparticles as delivery vehicles for osteosarcoma. Nanomedicine Nanotechnol Biol Med. 2012 May;8(4):440–451.

120. Škalko N, Brandl M, Bećirević-Laćan M, Filipović-Grčić J, Jalšenjak I. Liposomes with nifedipine and nifedipine-cyclodextrin complex: calorimetrical and plasma stability comparison. Eur J Pharm Sci. 1996 Nov;4(6):359–366.

121. Zhu Q, Guo T, Xia D, Li X, Zhu C, Li H, et al. Pluronic F127-modified liposome-containing tacrolimus-cyclodextrin inclusion complexes: improved solubility, cellular uptake and intestinal penetration. J Pharm Pharmacol. 2013 Aug;65(8):1107–1117.

122. Muheim C, Götzke H, Eriksson AU, Lindberg S, Lauritsen I, Nørholm MHH, et al. Increasing the permeability of Escherichia coli using MAC13243. Sci Rep. 2017 Dec;7(1):17629.

123. Komljenović I, Marquardt D, Harroun TA, Sternin E. Location of chlorhexidine in DMPC model membranes: a neutron diffraction study. Chem Phys Lipids. 2010 Jun;163(6):480–487.

124. Dickey A, Faller R. Examining the contributions of lipid shape and headgroup charge on bilayer behavior. Biophys J. 2008 Sep;95(6):2636–2646.

125. Jouhet J. Importance of the hexagonal lipid phase in biological membrane organization. Front Plant Sci. 2013;4:494.

126. Khalifat N, Fournier J-B, Angelova MI, Puff N. Lipid packing variations induced by pH in cardiolipin-containing bilayers: The driving force for the cristae-like shape instability. Biochim Biophys Acta BBA - Biomembr. 2011 Nov;1808(11):2724–2733.

127. White GF, Racher KI, Lipski A, Hallett FR, Wood JM. Physical properties of liposomes and proteoliposomes prepared from Escherichia coli polar lipids. Biochim Biophys Acta BBA - Biomembr. 2000 Sep;1468(1–2):175–186.

128. Brown RE, Brockman HL. Using Monomolecular Films to Characterize Lipid Lateral Interactions. Methods Mol Biol Clifton NJ. 2007;398:41–58.

129. Brockman H. Dipole potential of lipid membranes. Chem Phys Lipids. 1994 Sep;73(1–2):57–79.

130. Marsh D. Lateral pressure in membranes. Biochim Biophys Acta. 1996 Oct;1286(3):183–223.

131. Penfold J, Thomas RK. The application of the specular reflection of neutrons to the study of surfaces and interfaces. J Phys Condens Matter. 1990 Feb;2(6):1369–1412.

References

133

132. Doshi DA, Dattelbaum AM, Watkins EB, Brinker CJ, Swanson BI, Shreve AP, et al. Neutron Reflectivity Study of Lipid Membranes Assembled on Ordered Nanocomposite and Nanoporous Silica Thin Films. Langmuir. 2005;21(7):2865–2870.

133. Sivia, D. S. (St John’s College O. Elementary Scattering Theory: For X-ray and Neutron Users. 1st ed. Oxford University Press; 2011.

134. Lakey JH. Neutrons for biologists: a beginner’s guide, or why you should consider using neutrons. J R Soc Interface. 2009 Oct;6 Suppl 5(Suppl 5):S567–573.

135. Clifton LA, Neylon C, Lakey JH. Examining Protein–Lipid Complexes Using Neutron Scattering. In: Methods in molecular biology. 2013. p. 119–150.

136. Lu JR, Su TJ, Thomas RK. Structural Conformation of Bovine Serum Albumin Layers at the Air–Water Interface Studied by Neutron Reflection. J Colloid Interface Sci. 1999 May;213(2):426–437.

137. Leimgruber W, Stefanović V, Schenker F, Karr A, Berger J. Isolation and characterization of anthramycin, a new antitumor antibiotic. J Am Chem Soc. 1965 Dec;87(24):5791–5793.

138. Mantaj J, Jackson PJM, Karu K, Rahman KM, Thurston DE. Covalent Bonding of Pyrrolobenzodiazepines (PBDs) to Terminal Guanine Residues within Duplex and Hairpin DNA Fragments. PloS One. 2016;11(4):e0152303.

139. Tse H, Gu Q, Sze K-H, Chu IK, Kao RY-T, Lee K-C, et al. A tricyclic pyrrolobenzodiazepine produced by Klebsiella oxytoca is associated with cytotoxicity in antibiotic-associated hemorrhagic colitis. J Biol Chem. 2017 Sep;292(47):19503–19520.

140. Gerratana B. Biosynthesis, synthesis, and biological activities of pyrrolobenzodiazepines. Med Res Rev. 2012 Mar;32(2):254–93.

141. Basher MA, Rahman KM, Jackson PJM, Thurston DE, Fox KR. Sequence-selective binding of C8-conjugated pyrrolobenzodiazepines (PBDs) to DNA. Biophys Chem. 2017 Nov;230:53–61.

142. Mantaj J, Jackson PJM, Rahman KM, Thurston DE. From Anthramycin to Pyrrolobenzodiazepine (PBD)-Containing Antibody–Drug Conjugates (ADCs). Angew Chem - Int Ed. 2017;56(2):462–488.

143. Brucoli F, Guzman JD, Basher MA, Evangelopoulos D, McMahon E, Munshi T, et al. DNA sequence-selective C8-linked pyrrolobenzodiazepine–heterocyclic polyamide conjugates show anti-tubercular-specific activities. J Antibiot (Tokyo). 2016 Dec;69(12):843–9.

144. Brown DG, May-Dracka TL, Gagnon MM, Tommasi R. Trends and Exceptions of Physical Properties on Antibacterial Activity for Gram-Positive and Gram-Negative Pathogens. J Med Chem. 2014 Dec;57(23):10144–10161.

145. Upadhyay S, Kumar G. NMR and molecular modelling studies on the interaction of fluconazole with β-cyclodextrin. Chem Cent J. 2009 Aug;3(1):9.

146. Loftsson T, Jarho P, Másson M, Järvinen T. Cyclodextrins in drug delivery. Expert Opin Drug Deliv. 2005 Mar;2(2):335–351.

147. Loftsson T, Másson M. Evaluation of cyclodextrin solubilization of drugs. Int J Pharm. 2005;302:18–28.

148. Gabelica V, Galic N, Pauw E. On the specificity of cyclodextrin complexes detected by electrospray mass spectrometry. J Am Soc Mass Spectrom. 2002 Aug;13(8):946–953.

References

134

149. Correia I, Bezzenine N, Ronzani N, Platzer N, Beloeil J-C, Doan B-T. Study of inclusion complexes of acridine with β- and (2,6-di-O-methyl)-β-cyclodextrin by use of solubility diagrams and NMR spectroscopy. J Phys Org Chem. 2002 Sep;15(9):647–659.

150. Tolman JA, Williams RO. Advances in the pulmonary delivery of poorly water-soluble drugs: influence of solubilization on pharmacokinetic properties. Drug Dev Ind Pharm. 2010 Jan;36(1):1–30.

151. Valero M, Castiglione F, Mele A, da Silva MA, Grillo I, González-Gaitano G, et al. Competitive and Synergistic Interactions between Polymer Micelles, Drugs, and Cyclodextrins: The Importance of Drug Solubilization Locus. Langmuir ACS J Surf Colloids. 2016 13;32(49):13174–86.

152. Nigam S, Durocher G. Spectral and Photophysical Studies of Inclusion Complexes of Some Neutral 3H-Indoles and Their Cations and Anions with β-Cyclodextrin. J Phys Chem. 1996 Jan;100(17):7135–7142.

153. Peeters J, Neeskens P, Tollenaere JP, Van Remoortere P, Brewster ME. Characterization of the interaction of 2-hydroxypropyl-β-cyclodextrin with itraconazole at pH 2, 4, and 7. J Pharm Sci. 2002 Jun;91(6):1414–1422.

154. Chen A, Liu M, Dong L, Sun D. Study on the effect of solvent on the inclusion interaction of hydroxypropyl-β-cyclodextrin with three kinds of coumarins by phase solubility method. Fluid Phase Equilibria. 2013 Mar;341:42–47.

155. Loftsson T, Brewster ME. Cyclodextrins as Functional Excipients: Methods to Enhance Complexation Efficiency. J Pharm Sci. 2012 Sep;101(9):3019–3032.

156. Huang CY. Determination of binding stoichiometry by the continuous variation method: The job plot. Methods Enzymol. 1982 Jan;87:509–525.

157. Bertholet P, Gueders M, Dive G, Albert A, Barillaro V, Perly B, et al. The effect of cyclodextrins on the aqueous solubility of a new MMP inhibitor: phase solubility, 1 H-NMR spectroscopy and molecular modeling studies, preparation and stability study of nebulizable solutions. J Pharm Pharm Sci Publ Can Soc Pharm Sci Soc Can Sci Pharm. 2005 Jul;8(2):163–174.

158. Angiolini L, Agnes M, Cohen B, Yannakopoulou K, Douhal A. Formation, characterization and pH dependence of rifampicin: heptakis(2,6-di- O -methyl)-β-cyclodextrin complexes. Int J Pharm. 2017 Oct;531(2):668–675.

159. Hirayama F, Kurihara M, Horiuchi Y, Utsuki T, Uekama K, Yamasaki M. Preparation of heptakis(2,6-di-O-ethyl)-beta-cyclodextrin and its nuclear magnetic resonance spectroscopic characterization. Pharm Res. 1993 Feb;10(2):208–13.

160. Iza N, Guerrero-Martínez A, Tardajos G, Ortiz MJ, Palao E, Montoro T, et al. Using Inclusion Complexes with Cyclodextrins To Explore the Aggregation Behavior of a Ruthenium Metallosurfactant. Langmuir. 2015 Mar;31(9):2677–2688.

161. Poór M, Zand A, Szente L, Lemli B, Kunsági-Máté S, Poór M, et al. Interaction of α- and β-zearalenols with β-cyclodextrins. Molecules. 2017 Nov;22(11):1910.

162. Tewes F, Brillault J, Couet W, Olivier J-C. Formulation of rifampicin–cyclodextrin complexes for lung nebulization. J Controlled Release. 2008 Jul;129(2):93–99.

163. Chadha R, Saini A, Gupta S, Arora P, Thakur D, Jain DVS. Encapsulation of rifampicin by natural and modified β-cyclodextrins: characterization and thermodynamic parameters. J Incl Phenom Macrocycl Chem. 2010 Jun;67(1–2):109–116.

References

135

164. He D, Deng P, Yang L, Tan Q, Liu J, Yang M, et al. Molecular encapsulation of rifampicin as an inclusion complex of hydroxypropyl-β-cyclodextrin: Design; characterization and in vitro dissolution. Colloids Surf B Biointerfaces. 2013 Mar;103:580–585.

165. International Conference on Harmonization. Impurities in New Drug Products - Q3B (R2) [Internet]. 2006 p. 12. Report No.: June. Available from: http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Q3B(R2)+-+Impurities+in+New+Drug+Products#2

166. Dabkowski J, Zagórska I, Dabkowska M, Koczorowski Z, Trasatti S. Adsorption of DMSO at the free surface of water: surface excesses and surface potential shifts in the low concentration range. J Chem Soc Faraday Trans. 1996 Jan;92(20):3873–3878.

167. David S. Karpovich, Ray D. Adsorption of Dimethyl Sulfoxide to the Liquid/Vapor Interface of Water and the Thermochemistry of Transport across the Interface. J Phys Chem B. 1998;102(4):649–652.

168. Christian F. Wertz, Santore MM. Effect of Surface Hydrophobicity on Adsorption and Relaxation Kinetics of Albumin and Fibrinogen: Single-Species and Competitive Behavior. Langmuir. 2001;17(10):3006–3016.

169. Wang J, Matayoshi E. Solubility at the molecular level: development of a critical aggregation concentration (CAC) assay for estimating compound monomer solubility. Pharm Res. 2012 Jul;29(7):1745–54.

170. Bonev B, Hooper J, Parisot J. Principles of assessing bacterial susceptibility to antibiotics using the agar diffusion method. J Antimicrob Chemother. 2008 Mar;61(6):1295–1301.

171. Loftsson T. Non-inclusion cyclodextrin complexes [Internet]. US7,115,586B2, 2004. Available from: https://patentimages.storage.googleapis.com/6d/3b/97/875e2cd02c0c26/US7115586.pdf

172. Upadhye SB, Kulkarni SJ, Majumdar S, Avery MA, Gul W, ElSohly MA, et al. Preparation and characterization of inclusion complexes of a hemisuccinate ester prodrug of delta9-tetrahydrocannabinol with modified beta-cyclodextrins. AAPS PharmSciTech. 2010 Jun;11(2):509–17.

173. Ferreira DA, Ferreira AG, Vizzotto L, Federman Neto A, de Oliveira AG. Analysis of the molecular association of rifampicin with hydroxypropyl-β-cyclodextrin. Rev Bras Ciênc Farm. 2004 Mar;40(1):43–51.

174. Drulis-Kawa Z, Gubernator J, Dorotkiewicz-Jach A, Doroszkiewicz W, Kozubek A. In vitro antimicrobial activity of liposomal meropenem against Pseudomonas aeruginosa strains. Int J Pharm. 2006 Jun;315(1–2):59–66.

175. Loftsson T, Hreinsdóttir D, Másson M. The complexation efficiency. J Incl Phenom Macrocycl Chem. 2007 Mar;57(1–4):545–552.

176. Liu L, Guo Q-X. The Driving Forces in the Inclusion Complexation of Cyclodextrins. J Incl Phenom Macrocycl Chem. 2002;42(1–2):1–14.

177. Antimicrobial resistance (EARS-Net) - Annual Epidemiological Report for 2014 [Internet]. [cited 2018 Oct 6]. Available from: https://ecdc.europa.eu/en/publications-data/antimicrobial-resistance-ears-net-annual-epidemiological-report-2014

178. Peleg AY, Hooper DC. Hospital-Acquired Infections Due to Gram-Negative Bacteria. N Engl J Med. 2010 May;362(19):1804–1813.

References

136

179. Sachetelli S, Khalil H, Chen T, Beaulac C, Sénéchal S, Lagacé J. Demonstration of a fusion mechanism between a fluid bactericidal liposomal formulation and bacterial cells. Biochim Biophys Acta. 2000 Feb;1463(2):254–66.

180. Akbarzadeh A, Rezaei-Sadabady R, Davaran S, Joo SW, Zarghami N, Hanifehpour Y, et al. Liposome: classification, preparation, and applications. Nanoscale Res Lett. 2013 Feb;8(1):102.

181. Huang Z, London E. Effect of Cyclodextrin and Membrane Lipid Structure upon Cyclodextrin–Lipid Interaction. Langmuir. 2013 Nov;29(47):14631–14638.

182. Anderson TG, Tan A, Ganz P, Seelig J. Calorimetric measurement of phospholipid interaction with methyl-beta-cyclodextrin. Biochemistry. 2004 Mar;43(8):2251–2261.

183. Liossi ΑS, Ntountaniotis D, Kellici TF, Chatziathanasiadou MV, Megariotis G, Mania M, et al. Exploring the interactions of irbesartan and irbesartan–2-hydroxypropyl-β-cyclodextrin complex with model membranes. Biochim Biophys Acta BBA - Biomembr. 2017 Jun;1859(6):1089–1098.

184. Hudzicki J. Kirby-Bauer Disk Diffusion Susceptibility Test Protocol. Am Soc Microbiol. 2009 Dec;1–23.

185. Zhang H-M, Li Z, Uematsu K, Kobayashi T, Horikoshi K. Antibacterial activity of cyclodextrins against Bacillus strains. Arch Microbiol. 2008 Nov;190(5):605–609.

186. Szoka F, Papahadjopoulos D. Comparative Properties and Methods of Preparation of Lipid Vesicles (Liposomes). Annu Rev Biophys Bioeng. 1980 Jun;9(1):467–508.

187. Bhattacharjee S. DLS and zeta potential – What they are and what they are not? J Controlled Release. 2016 Aug;235:337–351.

188. Stewart JCM. Colorimetric determination of phospholipids with ammonium ferrothiocyanate. Anal Biochem. 1980 May;104(1):10–14.

189. Bozzuto G, Molinari A. Liposomes as nanomedical devices. Int J Nanomedicine. 2015 Feb;10:975–999.

190. Burgess S. Liposome Preparation - Avanti® Polar Lipids | Sigma-Aldrich [Internet]. 1998 [cited 2018 Oct 7]. Available from: https://www.sigmaaldrich.com/technical-documents/articles/biology/liposome-preparation.html

191. Wetterau JR, Jonas A. Effect of dipalmitoylphosphatidylcholine vesicle curvature on the reaction with human apolipoprotein A-I. J Biol Chem. 1982 Sep;257(18):10961–10966.

192. Huang C-M, Chen C-H, Pornpattananangkul D, Zhang L, Chan M, Hsieh M-F, et al. Eradication of drug resistant Staphylococcus aureus by liposomal oleic acids. Biomaterials. 2011 Jan;32(1):214–221.

193. Hope MJ, Bally MB, Mayer LD, Janoff AS, Cullis PR. Generation of multilamellar and unilamellar phospholipid vesicles. Chem Phys Lipids. 1986 Jun;40(2–4):89–107.

194. Connor J, Sullivan S, Huang H L. Monoclonal Antibody And Liposomes. Pharmacol Ther. 1985;28(3):341–365.

195. Drulis-Kawa Z, Dorotkiewicz-Jach A. Liposomes as delivery systems for antibiotics. Int J Pharm. 2010 Mar;387(1–2):187–198.

References

137

196. Drulis-Kawa Z, Gubernator J, Dorotkiewicz-Jach A, Doroszkiewicz W, Kozubek A. A comparison of the in vitro antimicrobial activity of liposomes containing meropenem and gentamicin. Cell Mol Biol Lett. 2006;11(3):360–375.

197. Tandrup Schmidt S, Foged C, Smith Korsholm K, Rades T, Christensen D. Liposome-Based Adjuvants for Subunit Vaccines: Formulation Strategies for Subunit Antigens and Immunostimulators. Pharmaceutics. 2016 Mar;8(1):7.

198. Disalvo EA, Hollmann A, Martini MF. Hydration in Lipid Monolayers: Correlation of Water Activity and Surface Pressure. Subcell Biochem. 2015;71:213–231.

199. Disalvo EA, Lairion F, Martini F, Tymczyszyn E, Frías M, Almaleck H, et al. Structural and functional properties of hydration and confined water in membrane interfaces. Biochim Biophys Acta BBA - Biomembr. 2008 Dec;1778(12):2655–2670.

200. Binder H, Zschörnig O. The effect of metal cations on the phase behavior and hydration characteristics of phospholipid membranes. Chem Phys Lipids. 2002 May;115(1–2):39–61.

201. Gurtovenko AA, Vattulainen I. Effect of NaCl and KCl on Phosphatidylcholine and Phosphatidylethanolamine Lipid Membranes: Insight from Atomic-Scale Simulations for Understanding Salt-Induced Effects in the Plasma Membrane. J Phys Chem B. 2008 Feb;112(7):1953–1962.

202. Melcrová A, Pokorna S, Pullanchery S, Kohagen M, Jurkiewicz P, Hof M, et al. The complex nature of calcium cation interactions with phospholipid bilayers. Sci Rep. 2016 Dec;6(1):38035.

203. Szente L, Fenyvesi É. Cyclodextrin-Lipid Complexes: Cavity Size Matters. Struct Chem. 2017 Apr;28(2):479–492.

204. Szejtli J, Cserháti T, Szögyi M. Interactions between cyclodextrins and cell-membrane phospholipids. Carbohydr Polym. 1986 Jan;6(1):35–49.

205. Nishijo J, Shiota S, Mazima K, Inoue Y, Mizuno H, Yoshida J. Interactions of cyclodextrins with dipalmitoyl, distearoyl, and dimyristoyl phosphatidyl choline liposomes. A study by leakage of carboxyfluorescein in inner aqueous phase of unilamellar liposomes. Chem Pharm Bull (Tokyo). 2000 Jan;48(1):48–52.

206. Uhríková D, Kučerka N, Teixeira J, Gordeliy V, Balgavý P. Structural changes in dipalmitoylphosphatidylcholine bilayer promoted by Ca2+ ions: a small-angle neutron scattering study. Chem Phys Lipids. 2008 Oct;155(2):80–89.

207. Pedersen UR, Leidy C, Westh P, Peters GH. The effect of calcium on the properties of charged phospholipid bilayers. Biochim Biophys Acta BBA - Biomembr. 2006 May;1758(5):573–582.

208. Lapinski MM, Castro-Forero A, Greiner AJ, Ofoli RY, Blanchard GJ. Comparison of Liposomes Formed by Sonication and Extrusion: Rotational and Translational Diffusion of an Embedded Chromophore. Langmuir. 2007 Nov;23(23):11677–11683.

209. Connors KA, Pendergast DD. Microscopic binding constants in cyclodextrin systems: complexation of .alpha.-cyclodextrin with sym-1,4-disubstituted benzenes. J Am Chem Soc. 1984 Nov;106(24):7607–7614.

210. Nishijo J, Moriyama S, Shiota S. Interactions of cholesterol with cyclodextrins in aqueous solution. Chem Pharm Bull (Tokyo). 2003 Nov;51(11):1253–1257.

References

138

211. Ruysschaert T, Marque A, Duteyrat J-L, Lesieur S, Winterhalter M, Fournier D. Liposome retention in size exclusion chromatography. BMC Biotechnol. 2005 May;5(1):11.

212. Yamamoto S, Kurihara H, Mutoh T, Xing X, Unno H. Cholesterol recovery from inclusion complex of β-cyclodextrin and cholesterol by aeration at elevated temperatures. Biochem Eng J. 2005 Feb;22(3):197–205.

213. Machut-Binkowski C, Hapiot F, Cecchelli R, Martin P, Monflier E. A versatile liposome/cyclodextrin supramolecular carrier for drug delivery through the blood-brain barrier. J Incl Phenom Macrocycl Chem. 2007 Mar;57(1–4):567–572.

214. Donova MV, Nikolayeva VM, Dovbnya DV, Gulevskaya SA, Suzina NE. Methyl-beta-cyclodextrin alters growth, activity and cell envelope features of sterol-transforming mycobacteria. Microbiology. 2007 Jun;153(6):1981–1992.

215. Suárez DF, Consuegra J, Trajano VC, Gontijo SML, Guimarães PPG, Cortés ME, et al. Structural and thermodynamic characterization of doxycycline/β-cyclodextrin supramolecular complex and its bacterial membrane interactions. Colloids Surf B Biointerfaces. 2014 Jun;118:194–201.

216. Rosa Teixeira KI, Araújo PV, Almeida Neves BR, Bohorquez Mahecha GA, Sinisterra RD, Cortés ME. Ultrastructural changes in bacterial membranes induced by nano-assemblies β-cyclodextrin chlorhexidine: SEM, AFM, and TEM evaluation. Pharm Dev Technol. 2013 Jun;18(3):600–608.

217. Li X, Zhou L, Takai H, Sasaki Y, Mezawa M, Li Z, et al. Aggregatibacter actinomycetemcomitans lipopolysaccharide regulates bone sialoprotein gene transcription. J Cell Biochem. 2012 Sep;113(9):2822–2834.

218. Santos RS, Figueiredo C, Azevedo NF, Braeckmans K, De Smedt SC. Nanomaterials and molecular transporters to overcome the bacterial envelope barrier: Towards advanced delivery of antibiotics. Adv Drug Deliv Rev. 2017 Dec;136–137:28–48.

219. Jeworrek C, Evers F, Howe J, Brandenburg K, Tolan M, Winter R. Effects of Specific versus Nonspecific Ionic Interactions on the Structure and Lateral Organization of Lipopolysaccharides. Biophys J. 2011 May;100(9):2169–2177.

220. Le Brun AP, Clifton LA, Halbert CE, Lin B, Meron M, Holden PJ, et al. Structural characterization of a model gram-negative bacterial surface using lipopolysaccharides from rough strains of Escherichia coli. Biomacromolecules. 2013 Jun;14(6):2014–2022.

221. Santos NC, Silva AC, Castanho MARB, Martins-Silva J, Saldanha C. Evaluation of Lipopolysaccharide Aggregation by Light Scattering Spectroscopy. ChemBioChem. 2003 Jan;4(1):96–100.

222. Furse S, Scott DJ. Three-Dimensional Distribution of Phospholipids in Gram Negative Bacteria. Biochemistry. 2016 Aug;55(34):4742–4747.

223. Sohlenkamp C, Geiger O. Bacterial membrane lipids: diversity in structures and pathways. Narberhaus F, editor. FEMS Microbiol Rev. 2016 Jan;40(1):133–159.

224. Marquardt D, Geier B, Pabst G. Asymmetric Lipid Membranes: Towards More Realistic Model Systems. Membranes. 2015 May;5(2):180–196.

225. Clifton LA, Skoda MWA, Le Brun AP, Ciesielski F, Kuzmenko I, Holt SA, et al. Effect of Divalent Cation Removal on the Structure of Gram-Negative Bacterial Outer Membrane Models. Langmuir. 2015 Jan;31(1):404–412.

References

139

226. Clifton LukeA, Ciesielski F, Skoda MWA, Paracini N, Holt SA, Lakey JH. The Effect of Lipopolysaccharide Core Oligosaccharide Size on the Electrostatic Binding of Antimicrobial Proteins to Models of the Gram Negative Bacterial Outer Membrane. Langmuir. 2016 Apr;32(14):3485–3494.

227. Clifton LA, Holt SA, Hughes AV, Daulton EL, Arunmanee W, Heinrich F, et al. An Accurate In Vitro Model of the E. coli Envelope. Angew Chem Int Ed. 2015 Oct;54(41):11952–11955.

228. Clausell A, Garcia-Subirats M, Pujol M, Busquets MA, Rabanal F, Cajal Y. Gram-negative outer and inner membrane models: insertion of cyclic cationic lipopeptides. J Phys Chem B. 2007 Jan;111(3):551–563.

229. Vollhardt D, Fainerman VB. Progress in characterization of Langmuir monolayers by consideration of compressibility. Adv Colloid Interface Sci. 2006 Nov;127(2):83–97.

230. Dynarowicz-Łatka P, Hac-Wydro K. Interactions between phosphatidylcholines and cholesterol in monolayers at the air/water interface. Colloids Surf B Biointerfaces. 2004 Aug;37(1–2):21–5.

231. Hughes AV. RasCAL download | SourceForge.net [Internet]. 2013 [cited 2018 Oct 7]. Available from: https://sourceforge.net/projects/rscl/

232. Born M, Wolf Emil. Principles of optics : electromagnetic theory of propagation, interference and diffraction of light. Cambridge University Press; 1999. 952 p.

233. Haario H, Laine M, Mira A, Saksman E. DRAM: Efficient adaptive MCMC. Stat Comput. 2006 Dec;16(4):339–354.

234. Haario H, Saksman E, Tamminen J. An Adaptive Metropolis Algorithm. Bernoulli. 2001 Apr;7(2):223.

235. Duncan SL, Larson RG. Comparing Experimental and Simulated Pressure-Area Isotherms for DPPC. Biophys J. 2008 Apr;94(8):2965–2986.

236. Zuo YY, Chen R, Wang X, Yang J, Policova Z, Neumann AW. Phase Transitions in Dipalmitoylphosphatidylcholine Monolayers. Langmuir. 2016 Aug;32(33):8501–8506.

237. Shapovalov VL. Interaction of DPPC monolayer at air–water interface with hydrophobic ions. Thin Solid Films. 1998 Aug;327–329:599–602.

238. Dahmen-Levison U, Brezesinski G, Möhwald H. Specific adsorption of PLA2 at monolayers. Thin Solid Films. 1998 Aug;327–329:616–620.

239. Campbell RA, Saaka Y, Shao Y, Gerelli Y, Cubitt R, Nazaruk E, et al. Structure of surfactant and phospholipid monolayers at the air/water interface modeled from neutron reflectivity data. J Colloid Interface Sci. 2018 Dec;531:98–108.

240. Harvey RD, Ara N, Heenan RK, Barlow DJ, Quinn PJ, Lawrence MJ. Stabilization of Distearoylphosphatidylcholine Lamellar Phases in Propylene Glycol Using Cholesterol. Mol Pharm. 2013 Dec;10(12):4408–4417.

241. Galdston M, Shah DO. Surface properties and hysteresis of dipalmitoyllecithin in relation to the alveolar lining layer. Biochim Biophys Acta BBA - Lipids Lipid Metab. 1967 Apr;137(2):255–263.

242. García-Verdugo I, Cañadas O, Taneva SG, Keough KMW, Casals C. Surfactant Protein A Forms Extensive Lattice-Like Structures on 1,2-Dipalmitoylphosphatidylcholine/Rough-Lipopolysaccharide- Mixed Monolayers. Biophys J. 2007 Nov;93(10):3529–3540.

140

243. Herrmann M, Schneck E, Gutsmann T, Brandenburg K, Tanaka M. Bacterial lipopolysaccharides form physically cross-linked, two-dimensional gels in the presence of divalent cations. Soft Matter. 2015 Jul;11(30):6037–6044.

244. Michel JP, Wang YX, Dé E, Fontaine P, Goldmann M, Rosilio V. Charge and aggregation pattern govern the interaction of plasticins with LPS monolayers mimicking the external leaflet of the outer membrane of Gram-negative bacteria. Biochim Biophys Acta BBA - Biomembr. 2015 Nov;1848(11):2967–2979.

245. Lee H, Wu W, Oh JK, Mueller L, Sherwood G, Peteanu L, et al. Light-Induced Reversible Formation of Polymeric Micelles. Angew Chem Int Ed. 2007 Mar;46(14):2453–2457.

246. Athanassiou G, Michaleas S, Lada-Chitiroglou E, Tsitsa T, Antoniadou-Vyza E. Antimicrobial activity of β-lactam antibiotics against clinical pathogens after molecular inclusion in several cyclodextrins. A novel approach to bacterial resistance. J Pharm Pharmacol. 2003 Mar;55(3):291–300.

247. Bhamidimarri SP, Prajapati JD, van den Berg B, Winterhalter M, Kleinekathöfer U. Role of Electroosmosis in the Permeation of Neutral Molecules: CymA and Cyclodextrin as an Example. Biophys J. 2016 Feb;110(3):600–611.

248. Charitat T, Lecuyer S, Fragneto G. Fluctuations and destabilization of single phospholipid bilayers. Biointerphases. 2008 Jun;3(2):FB3–FB15.

249. Wang Z, Ma Y, Khalil H, Wang R, Lu T, Zhao W, et al. Fusion between fluid liposomes and intact bacteria: study of driving parameters and in vitro bactericidal efficacy. Int J Nanomedicine. 2016;11:4025–4036.

250. Los DA, Murata N. Membrane fluidity and its roles in the perception of environmental signals. Biochim Biophys Acta BBA - Biomembr. 2004 Nov;1666(1–2):142–157.

251. Chow AHL, Hsia CK, Gordon JD, Young JWM, Vargha-Butler EI. Assessment of wettability and its relationship to the intrinsic dissolution rate of doped phenytoin crystals. Int J Pharm. 1995 Dec;126(1–2):21–28.

252. Desjardins A, Chen T, Khalil H, Sayasith K, Lagacé J. Differential Behaviour of Fluid Liposomes Toward Mammalian Epithelial Cells and Bacteria: Restriction of Fusion to Bacteria. J Drug Target. 2002 Jan;10(1):47–54.

253. Coates A, Hu Y, Bax R, Page C. The future challenges facing the development of new antimicrobial drugs. Nat Rev Drug Discov. 2002 Nov;1(11):895–910.

254. Shen F, Chu S, Bence AK, Bailey B, Xue X, Erickson PA, et al. Quantitation of Doxorubicin Uptake, Efflux, and Modulation of Multidrug Resistance (MDR) in MDR Human Cancer Cells. J Pharmacol Exp Ther. 2007 Oct;324(1):95–102.

255. Loh B, Grant C, Hancock RE. Use of the fluorescent probe 1-N-phenylnaphthylamine to study the interactions of aminoglycoside antibiotics with the outer membrane of Pseudomonas aeruginosa. Antimicrob Agents Chemother. 1984 Oct;26(4):546–51.

256. Hatty CR, Le Brun AP, Lake V, Clifton LA, Liu GJ, James M, et al. Investigating the interactions of the 18kDa translocator protein and its ligand PK11195 in planar lipid bilayers. Biochim Biophys Acta BBA - Biomembr. 2014 Mar 1;1838(3):1019–30.

Appendix A

141

Appendix A Quantum-mechanic rule

According to Planck’s hypothesis, an oscillating wave of frequency, ν, is directly proportional to a defined energy, E (’quantum’), via Planck’s constant, h (EquationA1. 1). The velocity of the

periodic waves is related to the frequency, f, and the wavelength, λ, according toEquationA1. 2.

Therefore, Planck’s equation is transformed as in Equation 1.3

𝐸 = ℎ𝑓 Equation A1. 1

𝑣 = 𝜆𝑓 Equation A1. 2

𝐸 = ℎ ÄB Equation A1. 3

De Broglie connected the particle properties (mass and velocity) with a length scale by

proposingEquationA1. 4.

𝜆 = Å� Equation A1. 4

, where p is momentum of the particle is defined by the vectorial factor between the mass mn

and velocity v of the neutron.

𝑝 = 𝑚I𝑣 Equation A1. 5

Taking into accountEquationA1. 4 and EquationA1. 5, the particle-like behaviour of the neutron

is linked to its kinetic energy E by the following equation:

𝐸 = V@𝑚I𝑣@

V@

ÅZ

_UBZ Equation A1. 6

The kinetic energy of a beam of neutrons is thus related to the wavelength of the periodic neutron

wave. The energy of the neutron is also dependent upon the temperature T by a simplified

relationship:

𝐸 = 𝑘Ç𝑇 Equation A1. 7

, in which k_B is the Boltzmann’s constant.

Appendix B

142

Appendix B Table B 1: Minimum Inhibitory concentrations (MIC) of PPA148 against Gram-negative bacteria

in the absence and presence of the efflux pump inhibitor PaβN.

Compound

MIC (μg/mL)

Acinetobacter baumannii K. pseumoniae P. aeruginosa

ATTCC 17978 AYE M6 NCTC

13368 PA01 NCTC 13437

PP-A148 2 2 0.25-0.5 16-32 128 16

PP-A148 with PAβN 0.125-0.5 0.125-0.5 0.125-0.25 2-8 4-8 0.25-2

Table B 2: Minimum Inhibitory concentrations (MIC) of PPA148 against Gram-positive bacteria in the absence and presence of the efflux pump inhibitor PaβN.

Compound

MIC (μg/mL)

VRE VSE MRSA MSSA

NCTC 12201

NCTC 12204

NCTC 775

EMRSA 15

EMRSA 16

ARCC 9144

PP-A148 0.25-0.5 0.01 1 2 2

0.5

PP-A148 with PAβN <0.125 <0.125 <0.125 <0.125 <0.125 <0.125

Scheme B 1: Complete synthetic route of PPA148.

Appendix B

143

Figure B 1: 13CNMR of intermediate compounds of the PBD core synthesis. The samples were suspended in deuterated chloroform.

Figure B 2: 13CNMR of intermediate compounds of the tail synthesis. The samples were suspended in deuterated DMSO.

Appendix B

144

Figure B 3: 13CNMR of the final compound, PPA148. The sample was suspended in deuterated chloroform.

Figure B 4: The linear regression (A) met the assumptions that the residuals follow a normal distribution (B) and their variance is constant (B).

Appendix C

145

Appendix C Table C 1: Number and Volume distribution of empty and loaded fluidosomes.

Empty

Loaded with PPA148-RAMEB complex

Documents

2019_Vandera_Kelli_Kalliopi_1... - King's Research Portal